[go: up one dir, main page]

CN114819319A - Electric power system scheduling model solving method and system containing renewable energy - Google Patents

Electric power system scheduling model solving method and system containing renewable energy Download PDF

Info

Publication number
CN114819319A
CN114819319A CN202210403184.0A CN202210403184A CN114819319A CN 114819319 A CN114819319 A CN 114819319A CN 202210403184 A CN202210403184 A CN 202210403184A CN 114819319 A CN114819319 A CN 114819319A
Authority
CN
China
Prior art keywords
scene
renewable energy
bundle
scenes
confidence
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202210403184.0A
Other languages
Chinese (zh)
Other versions
CN114819319B (en
Inventor
陈雁
孙艳
莫东
李秋文
凌武能
卓毅鑫
林洁
吴茵
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
China South Power Grid International Co ltd
Guangxi Power Grid Co Ltd
Original Assignee
China South Power Grid International Co ltd
Guangxi Power Grid Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by China South Power Grid International Co ltd, Guangxi Power Grid Co Ltd filed Critical China South Power Grid International Co ltd
Priority to CN202210403184.0A priority Critical patent/CN114819319B/en
Publication of CN114819319A publication Critical patent/CN114819319A/en
Application granted granted Critical
Publication of CN114819319B publication Critical patent/CN114819319B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G06F18/232—Non-hierarchical techniques
    • G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N7/00—Computing arrangements based on specific mathematical models
    • G06N7/01—Probabilistic graphical models, e.g. probabilistic networks
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G06Q10/06313—Resource planning in a project environment
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Energy or water supply
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J3/00—Circuit arrangements for AC mains or AC distribution networks
    • H02J3/38—Arrangements for parallely feeding a single network by two or more generators, converters or transformers
    • H02J3/46—Controlling of the sharing of output between the generators, converters, or transformers
    • H02J3/466—Scheduling the operation of the generators, e.g. connecting or disconnecting generators to meet a given demand
    • H02J2103/30—

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Human Resources & Organizations (AREA)
  • General Physics & Mathematics (AREA)
  • Economics (AREA)
  • Strategic Management (AREA)
  • Data Mining & Analysis (AREA)
  • General Business, Economics & Management (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • Tourism & Hospitality (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Software Systems (AREA)
  • Evolutionary Computation (AREA)
  • Mathematical Physics (AREA)
  • Game Theory and Decision Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Development Economics (AREA)
  • Operations Research (AREA)
  • Health & Medical Sciences (AREA)
  • Probability & Statistics with Applications (AREA)
  • Computing Systems (AREA)
  • Quality & Reliability (AREA)
  • Water Supply & Treatment (AREA)
  • General Health & Medical Sciences (AREA)
  • Mathematical Optimization (AREA)
  • Primary Health Care (AREA)
  • Algebra (AREA)
  • Public Health (AREA)
  • Computational Mathematics (AREA)
  • Biodiversity & Conservation Biology (AREA)
  • Evolutionary Biology (AREA)
  • Educational Administration (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Mathematical Analysis (AREA)

Abstract

本发明提供一种含可再生能源的电力系统调度模型求解方法及系统，首先建立基于可再生能源出力场景的联合机会约束求解模型，然后采用bundle方法对原始场景进行预处理，得到计算复杂性大大降低的bundle场景集，最后根据得到的bundle场景建立将联合机会约束转化为确定性约束的求解模型。本发明采用bundle的思想对其进行分析及预处理，将不影响其最终求解结果的大量相似场景整合成bundle束，进而减小模型的复杂程度，在不影响求解准确性的同时大大提升了联合机会约束的求解效率。

Figure 202210403184

The invention provides a method and system for solving a dispatching model of a power system including renewable energy. First, a joint chance constraint solving model based on a renewable energy output scenario is established, and then a bundle method is used to preprocess the original scenario, and the calculation complexity is greatly increased. After reducing the bundle scenario set, finally, according to the obtained bundle scenario, a solution model that converts the joint chance constraint into a deterministic constraint is established. The invention adopts the idea of bundles to analyze and preprocess them, and integrates a large number of similar scenes that do not affect the final solution result into bundles, thereby reducing the complexity of the model, and greatly improving the combination without affecting the accuracy of the solution. The solution efficiency of chance constraints.

Figure 202210403184

Description

Solving method and system for power system scheduling model containing renewable energy
Technical Field
The invention belongs to the field of power system scheduling, and particularly relates to a method and a system for solving a scheduling model of a power system containing renewable energy.
Background
The proposal of the double-carbon target puts higher requirements on the scheduling problem of large-proportion renewable energy grid connection, new energy such as wind power, photovoltaic and small hydropower stations have the characteristics of volatility and intermittency, and the problem to be solved is urgent to deal with the characteristic of good randomness in the scheduling problem of the power system without losing timeliness and accuracy. At present, a new energy power generation device often has the characteristics of wide distribution and strong randomness, so that new energy output of a multi-region power system is measured by adopting a multi-dimensional random variable, and stochastic programming is a powerful means for solving an optimization problem containing the random variable, a stochastic programming method capable of solving the optimization problem containing the multi-dimensional random variable is researched, and the method has important significance for safe and stable operation of the power system.
Stochastic programming requires the description of the probability distribution of random variables with a certain number of typical scenarios. The opportunity constraint is an important branch of random planning, belongs to nested probability constraint, and cannot be directly solved by adopting a conventional deterministic constraint method. The opportunity constraint is divided into a joint opportunity constraint and a single opportunity constraint, the joint opportunity constraint is complex in form and difficult to solve, the joint opportunity constraint is converted into a deterministic constraint before the joint opportunity constraint is solved, and the conversion method is divided into an analog conversion method and an analytic conversion method. The main idea of solving the opportunity constraint problem by the simulation method is to generate a plurality of samples of random variables to form a sample set on the basis of Monte Carlo sampling, approximate the probability distribution of the random variables to the probability distribution of elements in the sample set, and replace the random variables in the opportunity constraint with typical scene values corresponding to confidence degrees required by the opportunity constraint. Sample Average Approximation (SAA) is the most typical method of simulation. In order to solve the randomness problem of wind power, a document A micro-computer energy management system based on random-constrained stored-stochastic optimization and big data analysis proposes an opportunity-constrained two-stage stochastic programming optimization model, and combines a combined SAA algorithm to solve a model feasible solution. Although the simulation method is widely applied and easy to implement, the deterministic constraint obtained by the simulation method has the defects of long calculation time, low fitting precision and the like, and the problem of joint opportunity constraint containing multidimensional random variables still needs to be continuously explored.
In order to solve the defects of the simulation method, a scholars provides an analytic transformation method, and the main idea of the analytic transformation method is to analyze and reconstruct a probability distribution function of a random variable through a mathematical method and directly transform an opportunity constraint into a deterministic constraint based on the constructed probability distribution function. The traditional analytic transformation method assumes that the probability distribution of random variables obeys known probability distribution, for example, the document "Multi-objective distributed generation planning in distribution network configuration relating to computing environments", proposes an opportunity constraint model, assumes that the wind speed obeys weibull distribution, the photovoltaic output obeys beta distribution, and the load uncertainty obeys gaussian distribution. However, this assumption is not accurate enough and the actual random variables do not always obey some known probability distribution. The document "adaptive Robust route sequence-Constrained MILP Model for multistate Distribution System Planning with Uncertain Renewables and Loads" proposes a reconstruction method for distributed Robust optimization, which converts a single-opportunity constraint Model containing a single-dimensional random variable into a two-stage optimization problem and then into a deterministic optimization problem by assuming that the random variable obeys some possible distributions, but when the random variable is high-dimensional, the transformation process of the reconstruction method becomes very complicated and cannot necessarily be converted into a solvable convex problem. Patent document CN109728578A, "power system random dynamic unit combination method for solving quantiles based on newton method", adopts newton method to iteratively solve multidimensional random variables represented by opportunity constraint in a power system, but the iterative process seriously affects the rapidity of the solution, so that some data sets cannot obtain an effective solution in a short time. Although the method can solve the problem of partial opportunity constraint, the traditional joint opportunity constraint solving method is large in calculation amount and long in time consumption, and the requirement for quick calculation of power system scheduling personnel can not be met.
In summary, the existing method for solving the scheduling problem of the power system including renewable energy still has a deficiency in considering the correlation of renewable energy, and the traditional solution method based on joint opportunity constraint has a large calculation amount and a long time consumption, which brings inconvenience to the scheduling staff, so that a reliable, accurate and rapid scheduling method of the power system including renewable energy is urgently needed.
Disclosure of Invention
Aiming at the defects of the prior art, the invention aims to provide a method and a system for solving a power system scheduling model containing renewable energy, and aims to solve the problems that the existing joint opportunity constraint method is long in time consumption when solving a nonlinear mixed integer model with more dimensions, and brings inconvenience to scheduling personnel when being applied to actual power system scheduling.
In order to achieve the above object, in a first aspect, the present invention provides a method for solving a scheduling model of a power system including renewable energy, including the following steps:
the output of renewable energy sources is taken as a random variable, and joint opportunity constraint of the power system is established; the power system comprises a plurality of renewable energy sources, the random variable is multidimensional, and each dimension of data corresponds to the output of one renewable energy source; the joint opportunity constraint is a constraint relation in a multi-region power system scheduling model, and the power system scheduling model determines scheduling schemes of thermal power generating units and tie lines in each time period by taking the minimum total operating cost of a power system as a target;
determining historical data of corresponding random variables according to historical data of renewable energy output; the renewable energy output corresponding to one historical data of the random variable is used as a scene, the probability of each scene is determined according to the repeated condition of each scene in all the scenes, and the set of different scenes is used as the scene set of the random variable;
taking historical data of the random variable as a sample, and determining the cumulative distribution function of the random variable by adopting self-adaptive kernel density estimation based on the sample of the random variable;
determining an inverse function of the corresponding accumulative distribution function based on the accumulative distribution function of the renewable energy output, and substituting the single-opportunity constraint confidence coefficient into the inverse function to determine a confidence coefficient quantile point; the single opportunity constraint confidence is determined from a risk level of the joint opportunity constraint;
based on the confidence quantile, dividing an approximate scene which does not influence the final result of the mixed integer programming solution in the scene set and an effective scene to obtain the scene set of which each dimension of random variables is subjected to bundle processing; synthesizing the approximate scenes in the processed scene set into a bundle, so that the bundle and the effective scenes in the processed scene set jointly form a bundle scene set of random variables;
solving to obtain a target effective scene value meeting the confidence coefficient requirement by taking the minimum sum of all dimensional scene values of a bundle scene set generated by the output of renewable energy as a target function;
and converting the joint opportunity constraint of the power system into a deterministic constraint by using the target effective scene value, and solving a power system scheduling model corresponding to the deterministic constraint.
Further, the method for establishing the joint opportunity constraint of the power system by using the output of the renewable energy as a random variable specifically comprises the following steps:
the general expression of the joint opportunity constraint is:
Figure BDA0003601142830000041
P G a vector representing the output of the thermal power generating unit, y represents a vector corresponding to a parameter other than the renewable energy output and the thermal power generating unit output, P R Is an uncertain renewable energy output variable, which is a K-dimensional random variable, g k (P G Y) is a constraint equation containing the output of the thermal power generating unit, g k (P G ,y)≥-P R To constrain events, (1- α) JCC ) Is the joint opportunity constraint confidence, α JCC Is the risk level of the joint opportunity constraint, Pr {. represents the probability of satisfaction of the constraint event, and the joint opportunity constraint requires that the probability of satisfaction of K inequalities inside Pr {. is greater than or equal to (1-alpha) JCC )。
Further, the scene set of the random variables is determined by the following steps:
acquiring or updating the historical data of the output of the renewable energy sources, and acquiring a scene set capable of representing the randomness of the output of the renewable energy sources according to the historical data; let K dimension random variable history data
Figure BDA0003601142830000042
The resulting limited set of scenes is C,
Figure BDA0003601142830000043
the method is a scene in the historical data of the output of the renewable energy, wherein si belongs to C,
Figure BDA0003601142830000044
performing statistical clustering on the same scene in the scene set for the output scene value of the kth-dimension renewable energy in the sih scene to obtain a new scene set S, wherein S belongs to S,
Figure BDA0003601142830000045
for a new scenario in the renewable energy contribution history, wherein
Figure BDA0003601142830000046
The output scene value of the kth dimension renewable energy source of the S scene in the new scene set S, wherein the S scene comprises an initial scene d si Is M s ；
If the total number of scenes in C is Sn, then scene d in S s Probability P of s Comprises the following steps: p s ＝M s /Sn s∈S；
Figure BDA0003601142830000051
And acquiring a scene set S generated by the renewable energy output historical data and the corresponding probability of each scene in the scene set S.
Further, the sample based on the random variable determines the cumulative distribution function of the random variable by adopting adaptive kernel density estimation, specifically:
calculating the fixed bandwidth of the adaptive kernel density estimation by adopting the principle of minimum mean integral square error;
calculating cumulative distribution function of renewable energy output based on the fixed bandwidth
Figure BDA0003601142830000052
Figure BDA0003601142830000053
Wherein, N is the number of samples,
Figure BDA0003601142830000054
is the kernel function, H is the fixed bandwidth under the kernel function,
Figure BDA0003601142830000055
cumulative distribution function for fixed bandwidth, x is an independent variable representing renewable energy output, x i The ith sample value of the renewable energy output is obtained;
dividing N samples into M intervals, continuously iterating the intervals in an increasing mode from small to large, and determining the number of the intervals by judging whether the fitting error is smaller than an error threshold value;
calculating a position parameter for each interval
Figure BDA0003601142830000056
Wherein, the j, l interval is numbered,
Figure BDA0003601142830000057
an abscissa corresponding to the median of the sample cumulative distribution of the interval l; the sample bandwidth of each interval becomes H × H j Beta is the sensitivity coefficient, 0<β<1；
Self-adaptive bandwidth kernel density estimation is carried out by utilizing the position parameter of each interval, and the probability density function of the output of the renewable energy is calculated
Figure BDA0003601142830000058
Wherein, N j I is the number of samples in the interval j, i is the number of sample points in the interval j,
Figure BDA0003601142830000059
computing probability density functions for kernel estimation
Figure BDA00036011428300000510
With the true distribution f (x) i ) Mean square error MSE of (a):
Figure BDA00036011428300000511
the larger the partition number M is, the finer the bandwidth correction is, and the lower the error of the overall fitting is; by setting the error threshold MSE T When MSE>MSE R Then, the number M of the intervals is continuously increased, the step length of the intervals is increased, and the position parameters and the probability density function are repeatedly calculated until MSE<MSE T Up to or up to the maximumThe number of partitions is the maximum number of partitions is the number of samples, and the number of intervals obtained at the moment is used as the number of intervals which are determined finally;
after the interval number is finally determined, the renewable energy output P is calculated by utilizing interval self-adaptive bandwidth kernel density estimation R The cumulative distribution function of (1) is written as
Figure BDA0003601142830000061
Figure BDA0003601142830000062
x k And outputting the k-dimension renewable energy.
Further, the bundle scene set of the random variable is determined by the following steps:
calculating the inverse function of the cumulative distribution function of the renewable energy output of each dimension, and recording as
Figure BDA0003601142830000063
Constraining each single opportunity to confidence 1-alpha ICC Substituting into the inverse function of the cumulative distribution function to obtain confidence quantile points
Figure BDA0003601142830000064
Quantile based on confidence
Figure BDA0003601142830000065
Dividing an approximate scene which does not influence the final result of the mixed integer programming solution in the original scene set and an effective scene, and respectively processing each dimension random variable, wherein the processing method comprises the following steps:
Figure BDA0003601142830000066
thus, the s-th scene of each dimension of random variable after bundle preprocessing is obtained;
after the scenes of all the dimensional random variables are divided, a plurality of approximate scenes before a confidence quantile point and an effective scene set after the confidence quantile point are formed, and the approximate scenes are gathered into a new bundle, wherein the method comprises the following steps:
Figure BDA0003601142830000067
wherein s is b Representing bundle of renewable energy output scenes, wherein Sn is the total number of scenes of the original scene set, s e Representing the effective scene set of renewable energy output, and the two jointly form a new bundle scene set s B Each scene having a probability of
Figure BDA0003601142830000068
Wherein,
Figure BDA0003601142830000069
representing an ith scene of the bundle scene set, wherein each scene comprises k-dimensional renewable energy output scene values; bn represents the number of scenes constituting a bundle; bn represents the number of scenes that will newly generate the bundle scene set, and Sn is Bn + Bn-1.
Further, the solving to obtain a target effective scene set meeting the confidence requirement specifically includes:
the objective function is set up as follows:
Figure BDA0003601142830000071
the objective function requires that the sum of the dimensions of the scene generated by the renewable energy output is minimal,
Figure BDA0003601142830000072
is the k-dimension target scene value;
the constraints of the above objective function are as follows:
Figure BDA0003601142830000073
representing a kth dimension scene value in the ith scene of the Bundle scene set, and requiring that a target scene value is larger than any Bundle scene value meeting the condition;
Figure BDA0003601142830000074
the sum of the corresponding probabilities of all scenes in the scene set is required to be 1;
Figure BDA0003601142830000075
is a binary variable, and
Figure BDA0003601142830000076
with historical scenes
Figure BDA0003601142830000077
Correspondingly, each history scene
Figure BDA0003601142830000078
All have corresponding thereto
Figure BDA0003601142830000079
Requiring the ith target value to be greater than the product of the binary variable and the historical scene;
substituting the target function into a solver to solve based on the constraint condition to obtain a target effective scene value meeting the confidence coefficient requirement
Figure BDA00036011428300000710
Further, the solving of the power system model specifically includes:
the target effective scene value can satisfy the condition that the joint opportunity constraint is established under a certain confidence coefficient condition, so that an uncertain model corresponding to the original joint opportunity constraint is converted into a deterministic model, wherein the obtained target scene value is utilized
Figure BDA00036011428300000711
The resulting transformation opportunity constraints are:
Figure BDA00036011428300000712
and solving the deterministic model by using a solver to obtain the optimal value of each decision variable.
In a second aspect, the present invention provides a power system scheduling model solving system including renewable energy, including:
the joint opportunity constraint model establishing unit is used for establishing joint opportunity constraint of the power system by taking the output of the renewable energy as a random variable; the power system comprises a plurality of renewable energy sources, the random variable is multidimensional, and each dimension of data corresponds to the output of one renewable energy source; the joint opportunity constraint is a constraint relation in a multi-region power system scheduling model, and the power system scheduling model determines scheduling schemes of thermal power generating units and tie lines in each time period by taking the minimum total operating cost of a power system as a target;
the random variable scene set determining unit is used for determining historical data of corresponding random variables according to historical data of renewable energy output; the renewable energy output corresponding to one historical data of the random variable is used as a scene, the probability of each scene is determined according to the repeated condition of each scene in all the scenes, and the set of different scenes is used as the scene set of the random variable;
the cumulative distribution function determining unit is used for taking historical data of the random variable as a sample and determining the cumulative distribution function of the random variable by adopting self-adaptive kernel density estimation based on the sample of the random variable;
the confidence quantile determining unit is used for determining an inverse function of the corresponding accumulative distribution function based on the accumulative distribution function of the renewable energy output and substituting the single-opportunity constraint confidence into the inverse function to determine a confidence quantile; the single opportunity constraint confidence is determined from a risk level of the joint opportunity constraint;
a bundle scene set determining unit, configured to divide an approximate scene that does not affect a final result of mixed integer programming solution from an effective scene in the scene set based on the confidence quantile, to obtain a scene set in which the random variables of each dimension are subjected to bundle processing; synthesizing the approximate scenes in the processed scene set into a bundle, so that the bundle and the effective scenes in the processed scene set jointly form a bundle scene set of random variables;
the effective scene set determining unit is used for solving to obtain a target effective scene value meeting the confidence coefficient requirement by taking the minimum sum of all dimensional scene values of the bundle scene set generated by the renewable energy output as a target function;
and the deterministic model solving unit is used for converting the joint opportunity constraint of the power system into a deterministic constraint by using the target effective scene value and solving the power system scheduling model corresponding to the deterministic constraint.
Further, the bundle scene set determining unit determines a bundle scene set of the random variable by: calculating the inverse function of the cumulative distribution function of the renewable energy output of each dimension, and recording as
Figure BDA0003601142830000081
Constraining each single opportunity to confidence 1-alpha ICC Substituting into the inverse function of the cumulative distribution function to obtain confidence quantile points
Figure BDA0003601142830000082
Quantile based on confidence
Figure BDA0003601142830000091
Dividing an approximate scene which does not influence the final result of the mixed integer programming solution in the original scene set and an effective scene, and respectively processing each dimension random variable, wherein the processing method comprises the following steps:
Figure BDA0003601142830000092
thus, s scenes of random variables of each dimension after bundle preprocessing are obtained; after the scenes of the random variables of each dimension are divided, a plurality of approximate scenes before the confidence quantile and an effective scene set after the confidence quantile are formed, and the approximate scenes are gathered into a new bundle, wherein the method comprises the following steps:
Figure BDA0003601142830000093
wherein s is b Representing bundle of renewable energy output scenes, wherein Sn is the total number of scenes of the original scene set, s e Representing the effective scene set of renewable energy output, and the two jointly form a new bundle scene set s B Each scene having a probability of
Figure BDA0003601142830000094
Wherein,
Figure BDA0003601142830000095
representing an ith scene of the bundle scene set, wherein each scene comprises k-dimensional renewable energy output scene values; bn represents the number of scenes constituting a bundle; bn represents the number of scenes that will newly generate the bundle scene set, and Sn is Bn + Bn-1.
Further, the effective scene set determining unit solves to obtain a target effective scene set meeting the confidence requirement, specifically: the objective function is set up as follows:
Figure BDA0003601142830000096
the objective function requires that the sum of the dimensions of the scenario generated by the renewable energy output is minimal,
Figure BDA0003601142830000097
is the k-dimension target scene value; the constraints of the above objective function are as follows:
Figure BDA0003601142830000098
representing a kth dimension scene value in the ith scene of the Bundle scene set, and requiring that a target scene value is larger than any Bundle scene value meeting the condition;
Figure BDA0003601142830000099
the sum of the corresponding probabilities of all scenes in the scene set is required to be 1;
Figure BDA00036011428300000910
is a binary variable, and
Figure BDA00036011428300000911
with historical scenes
Figure BDA00036011428300000912
Correspondingly, each history scene
Figure BDA00036011428300000913
All have corresponding thereto
Figure BDA0003601142830000101
Requiring the ith target value to be greater than the product of the binary variable and the historical scene; substituting the target function into a solver to solve based on the constraint condition to obtain a target effective scene value meeting the confidence coefficient requirement
Figure BDA0003601142830000102
Generally, compared with the prior art, the above technical solution conceived by the present invention has the following beneficial effects:
the invention provides a method and a system for solving a power system dispatching model containing renewable energy sources.
The invention provides a method and a system for solving a scheduling model of a power system containing renewable energy, wherein the existing joint opportunity constraint method is long in time consumption when solving a nonlinear mixed integer model with more dimensions and brings great inconvenience to scheduling personnel when being applied to actual power system scheduling.
Drawings
Fig. 1 is a flowchart of a method for solving a scheduling model of a power system including renewable energy according to an embodiment of the present invention;
fig. 2 is a diagram of a system architecture for solving a scheduling model of a power system including renewable energy provided in an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
The proposal of the double-carbon target puts higher requirements on the scheduling problem of large-proportion renewable energy grid connection, new energy such as wind power, photovoltaic and small hydropower stations have the characteristics of volatility and intermittency, and the problem to be solved is urgent to deal with the characteristic of good randomness in the scheduling problem of the power system without losing timeliness and accuracy.
The existing method for processing the randomness of the renewable energy sources is long in time consumption or poor in accuracy. Aiming at the characteristic, the invention provides a method for improving settlement efficiency of a joint opportunity constraint scheduling model, and the method is used for improving the traditional joint opportunity constraint scheduling model. The method comprises the steps of firstly establishing a joint opportunity constraint solving model based on a renewable energy output scene, then preprocessing an original scene by adopting a bundle method to obtain a bundle scene set with greatly reduced computational complexity, and finally establishing a solving model for converting joint opportunity constraint into deterministic constraint according to the obtained bundle scene, so that the scheduling model containing the renewable energy is rapidly and accurately solved. The method is simple and practical, and can quickly and effectively process the scheduling problem of the power system.
The invention aims to provide a method for improving settlement efficiency of scheduling problems of a power system containing renewable energy sources based on joint opportunity constraint, aiming at the problem that the scheduling problems of the power system containing the renewable energy sources are long in solving time.
Fig. 1 is a flowchart of a method for solving a scheduling model of a power system including renewable energy according to an embodiment of the present invention; as shown in fig. 1, the method comprises the following steps:
s101, establishing joint opportunity constraint of a power system by taking the output of renewable energy as a random variable; the power system comprises a plurality of renewable energy sources, the random variable is multidimensional, and each dimension of data corresponds to the output of one renewable energy source;
s102, determining historical data of corresponding random variables according to historical data of renewable energy output; the renewable energy output corresponding to one historical data of the random variable is used as a scene, the probability of each scene is determined according to the repeated condition of each scene in all the scenes, and the set of different scenes is used as the scene set of the random variable;
s103, taking historical data of the random variable as a sample, and determining the cumulative distribution function of the random variable by adopting self-adaptive kernel density estimation based on the sample of the random variable;
s104, determining an inverse function of the corresponding accumulative distribution function based on the accumulative distribution function of the renewable energy output, and substituting the single-opportunity constraint confidence coefficient into the inverse function to determine a confidence coefficient quantile point; the single opportunity constraint confidence is determined from a risk level of the joint opportunity constraint;
s105, based on the confidence quantile, dividing an approximate scene which does not influence the final result of the mixed integer programming solution in the scene set and an effective scene to obtain the scene set of which each dimension random variable is subjected to bundle processing; synthesizing the approximate scenes in the processed scene set into a bundle so that the bundle and the effective scenes in the processed scene set jointly form a bundle scene set of random variables;
s106, solving to obtain a target effective scene value meeting the confidence coefficient requirement by taking the minimum sum of all dimensional scene values of the bundle scene set generated by the output of the renewable energy as a target function;
s107, converting the joint opportunity constraint of the power system into a deterministic constraint by using the target effective scene value, and solving the deterministic model to realize the solution of the power system scheduling model.
Specifically, the power system scheduling model is composed of three parts, namely an objective function, a conventional constraint and a joint opportunity constraint.
Firstly, establishing an objective function of a power system scheduling model:
f＝min(f G +f S +f R +f L ) (1)
Figure BDA0003601142830000121
Figure BDA0003601142830000122
Figure BDA0003601142830000123
Figure BDA0003601142830000124
in the formula (f) G 、f S 、f R 、f L The method comprises the following steps of respectively representing the power generation cost, the starting and stopping cost, the standby cost and the tie line cost of the thermal power generating unit, wherein f represents the total running cost of a power system;
Figure BDA0003601142830000125
respectively representing secondary, primary and constant power generation cost coefficients of the thermoelectric generator group g in the region k;
Figure BDA0003601142830000131
the generated power of the thermoelectric generator group g in the area k at the moment t is shown;
Figure BDA0003601142830000132
representing the starting and stopping states of the thermoelectric generator group g in the area k at the moment t,
Figure BDA0003601142830000133
representing the starting and stopping states of the fire motor group g in the area k at the moment t-1;
Figure BDA0003601142830000134
starting and stopping cost of the thermoelectric generator group g in the region k;
Figure BDA0003601142830000135
spare capacity for region k at time t, C k Unit spare cost for region k, C R Total standby cost; lambda [ alpha ] l,k A cost factor representing the power transmitted by zone i to zone k,
Figure BDA0003601142830000136
the power transmitted from the tie line region to the region k through the link line at the moment t; lambda [ alpha ] k,l A cost factor representing the power transmitted by zone k to zone i,
Figure BDA0003601142830000137
the power transmitted to the region l through the region k via the tie line at the time t; t is the total time period, and the invention takes 24 hours a day; k is the total number of regions, G k For a collection of thermal power generating units in region k, A k Is the set of links associated with region k.
Secondly, solving conventional constraints of the model, wherein the conventional constraints comprise power balance constraints, thermal power unit output upper and lower limit constraints, thermal power unit start-stop time constraints, thermal power unit climbing power constraints, reserve capacity constraints and tie line exchange power constraints, and accordingly, establishing conventional constraint conditions as shown in the following formula:
(1) power balance constraint
Figure BDA0003601142830000138
In the formula,
Figure BDA0003601142830000139
for the total load of the region k at the time t,
Figure BDA00036011428300001310
the effort is predicted for the renewable energy source of region k at time t.
(2) Thermal power generating unit output upper and lower limit restraint
Figure BDA00036011428300001311
In the formula,
Figure BDA00036011428300001312
and
Figure BDA00036011428300001313
respectively representing the lower limit and the upper limit of the output of the k thermal power generating unit g in the region;
(3) thermal power unit start-stop time constraint
Figure BDA0003601142830000141
In the formula,
Figure BDA0003601142830000142
and
Figure BDA0003601142830000143
respectively representing the switched-on time and the switched-off time of the unit g of the zone k,
Figure BDA0003601142830000144
and
Figure BDA0003601142830000145
respectively the minimum start-up and shut-down time of the unit g of the region k.
(4) Ramp power constraint of thermal power generating unit
Figure BDA0003601142830000146
In the formula,
Figure BDA0003601142830000147
and
Figure BDA0003601142830000148
the maximum upward and downward climbing power of the unit g in the region k respectively.
(5) Spare capacity constraint
Figure BDA0003601142830000149
In the formula,
Figure BDA00036011428300001410
the reserve capacity of the unit g in the area k at the time t.
(6) Junctor exchange power constraint
Figure BDA00036011428300001411
Figure BDA00036011428300001412
In the formula,p l,k and
Figure BDA00036011428300001413
respectively an upper limit and a lower limit of the transmission power from the area l to the area k,
Figure BDA00036011428300001414
the power flowing from zone i to zone k for time t,
Figure BDA00036011428300001415
and
Figure BDA00036011428300001416
represents the transmission state of the connecting line from l to k direction and from k to l direction (taking 1 represents flowing, and 0 represents not flowing).
Third, establish a joint opportunity constraint including spinning reserve as follows:
Figure BDA0003601142830000151
Figure BDA0003601142830000152
Figure BDA0003601142830000153
in the formula, eta is the standby rate; epsilon is the degree of confidence that,
Figure BDA0003601142830000154
indicating the rotational reserve capacity of region k at time t.
In a specific embodiment, the method for achieving the above object comprises the following steps:
step 1: establishing joint opportunity constraints containing multidimensional random variables
Establishing a joint opportunity constraint general expression as shown in formula (1-1):
Figure BDA0003601142830000155
P G vector representing power of thermal power generating unit, and of formula (15)
Figure BDA0003601142830000156
Corresponding to y representing the vector of the other parameter variable, and in equation (15)
Figure BDA0003601142830000157
η, etc. corresponds to P R For uncertain renewable energy output variables, as k-dimensional random variables, and in equation (15)
Figure BDA0003601142830000158
Corresponds to, g k (P G ,y)≥-P R To constrain events, α JCC Is the risk level of joint opportunity constraint, 1-alpha JCC Is the joint opportunity constraint confidence, corresponding to ε in equation (15), and Pr {. cndot.) represents the probability of satisfaction of the constraint event, equation (1-1) requires PrThe probability that K inequalities in {. is simultaneously satisfied is greater than or equal to 1-alpha JCC . Wherein y includes the magnitude of the load power
Figure BDA0003601142830000159
Rotating stand-by power
Figure BDA00036011428300001510
The utilization rate eta, etc.
Step 2: establishing scene set based on random variable historical data
Step 2.1 obtaining K-dimensional initial scene set
And acquiring or updating the historical data of the output of the renewable energy sources, and acquiring a scene set capable of representing the randomness of the output of the renewable energy sources according to the historical data. Let K dimension random variable history data
Figure BDA00036011428300001511
The resulting limited set of scenes is C,
Figure BDA00036011428300001512
for a scene in the renewable energy output historical data, wherein si belongs to C, the same scene in the scene set is subjected to statistical clustering to obtain a new scene set S, S belongs to S,
Figure BDA0003601142830000161
is one scene in the new renewable energy output historical data, wherein the s scene comprises an initial scene d si Is M s 。
Step 2.2 calculate the probability of each scene in S from 2.1
If the total number of scenes in C is Sn, then scene d in S s Probability P of s Comprises the following steps:
P s ＝M s /Sn s∈S (2-1)
Figure BDA0003601142830000162
the original scene generated from the renewable energy output historical data and its corresponding probability are thus obtained.
And step 3: and calculating the cumulative distribution function of the renewable energy output by using the historical data sample of the renewable energy output.
Step 3.1 computing the fixed Bandwidth of the Kernel Density estimate
The fixed bandwidth H is selected by using the principle of minimum mean-integral-square error, and when the kernel function selects the Gaussian function, the computational expression of the fixed bandwidth H is expressed by an expression (3-1), wherein
Figure BDA0003601142830000163
Is the standard deviation of the samples, and N is the number of samples of the random variable.
Figure BDA0003601142830000164
Step 3.2 calculate the cumulative distribution function of the renewable energy output by using the kernel density estimation of the fixed bandwidth
Calculating cumulative distribution function of renewable energy output by using formula (3-2)
Figure BDA0003601142830000165
Wherein, N is the number of samples,
Figure BDA0003601142830000166
for the kernel function, a gaussian function is generally selected as the kernel function, H is a fixed bandwidth under the gaussian kernel function,
Figure BDA0003601142830000167
a cumulative distribution function of a fixed bandwidth.
Step 3.3 determining number of intervals
And continuously iterating the interval number from small to large in an increasing mode, determining the interval number by judging whether the fitting error is smaller than an error threshold value, firstly setting the interval number M to be 5 for the first time, evenly distributing the samples to each interval if the N/M can be divided completely, and placing the samples of the remainder part which cannot be divided completely to the last interval if the N/M cannot be divided completely.
Calculating the position parameter h of each interval by using the formula (3-3) j ：
Figure BDA0003601142830000171
Wherein j, l is the number of the interval, M is the total number of the divided intervals,
Figure BDA0003601142830000172
a cumulative distribution function for a fixed bandwidth,
Figure BDA0003601142830000173
the abscissa corresponds to the median of the cumulative distribution of samples in the interval l. The sample bandwidth of each interval becomes H × H j Beta is the sensitivity coefficient, 0<β<1, when β is larger, h j The more sensitive, when β is equal to 0, the interval sample bandwidth becomes a fixed bandwidth H.
And calculating the probability density function of the output of the renewable energy source by using the position parameters and the interval adaptive bandwidth kernel density estimation of the formula (3-4).
Figure BDA0003601142830000174
Wherein N is j Is the number of samples in interval j, i is the number of sample points in interval j, N is the total number of samples in all intervals,
Figure BDA0003601142830000175
and (4) calculating the average square error of the probability density function of the kernel estimation and the real distribution, as shown in the formula (3-5).
Figure BDA0003601142830000176
The larger the number of partitions M, the finer the bandwidth correction,the lower the error of the global fit. By setting the error threshold MSE T It is possible to take the MSE T 60% -80% of the average square error at fixed bandwidth. When MSE>MSE T Then, the number M of the intervals is continuously increased, the step length of the intervals is increased to 5, and the probability density functions of the position parameters and the random variables are repeatedly calculated until MSE<MSE T Up to or up to a maximum number of partitions, which is the number of samples.
Step 3.4 calculating the cumulative distribution function of the random variables by using interval adaptive bandwidth kernel density estimation
After the interval number is determined, calculating the renewable energy output P by utilizing the interval adaptive bandwidth kernel density estimation of the formula (3-6) R Is recorded as the cumulative distribution function of
Figure BDA0003601142830000177
Figure BDA0003601142830000181
Wherein x is k And outputting the k-dimension renewable energy.
And 4, step 4: and generating a new bundle scene.
Step 4.1 of calculating confidence quantile points of renewable energy output
Calculating the inverse function of the cumulative distribution function of the renewable energy output of each dimension in the step 3.4, and recording the inverse function as
Figure BDA0003601142830000182
Constraining each single opportunity to confidence 1-alpha ICC Substituting into the inverse function of the cumulative distribution function to obtain confidence quantile points
Figure BDA0003601142830000183
Namely:
Figure BDA0003601142830000184
step 4.2, the original scene set is divided
Random variable confidence quantile obtained from step 4.1
Figure BDA0003601142830000185
Dividing an approximate scene which does not influence the final result of the mixed integer programming solution in the original scene set and an effective scene, and respectively processing each dimension random variable, wherein the processing method comprises the following steps:
Figure BDA0003601142830000186
therefore, s scenes of random variables of each dimension after bundle preprocessing are obtained.
Step 4.3, synthesizing the divided similar scenes into a bundle beam
After the scenes of the random variables of each dimension are divided, a plurality of approximate scenes before the quantile are formed and an effective scene set after the quantile are shown as a formula (4-3), and the approximate scenes are gathered into a new bundle, wherein the method comprises the following steps:
Figure BDA0003601142830000187
wherein s is b Bundle, s representing renewable energy output scenario e Representing the effective scene set of renewable energy output, and the two jointly form a new bundle scene set s B The probability calculation formula of each scene is as shown in formula (4-4)
Figure BDA0003601142830000191
Wherein,
Figure BDA0003601142830000192
representing the jth scene of the bundle scene set; bn represents the number of scenes constituting a bundle; bn represents the number of scenes that will newly generate the bundle scene set, and Sn is Bn + Bn-1.
And 5: and establishing a target scene solving model.
Step 5.1 objective function
The objective function is set up as follows:
Figure BDA0003601142830000193
the objective function requires a minimum sum of dimensions of the scene generated by the renewable energy output, wherein
Figure BDA0003601142830000194
Is the k-th dimension target scene value.
Step 5.2 constraint equation
The constraints for solving the model are as follows:
Figure BDA0003601142830000195
the constraint condition (5-2) requires that the target scene value is larger than any eligible Bundle scene value, wherein
Figure BDA0003601142830000196
Representing a k-dimension scene value in an ith scene of the bundle scene set.
Figure BDA0003601142830000197
And the constraint condition (5-3) requires that the sum of the corresponding probabilities of all scenes in the scene set is 1.
Figure BDA0003601142830000198
The constraint (5-4) indicates
Figure BDA0003601142830000199
Is a binary variable, and
Figure BDA00036011428300001910
with historical scenes
Figure BDA00036011428300001911
Correspondingly, each history scene
Figure BDA00036011428300001912
All have corresponding thereto
Figure BDA00036011428300001913
Figure BDA00036011428300001914
The constraint (5-5) requires that the jth target value be greater than the product of the binary variable and the historical scenario.
Step 5.3 solving the target scene model
Compared with the original scene, the number of scenes and the complexity of solution of the bundle scene set processed by the steps 4 and 5 are greatly reduced. Substituting the model established in the step 5 into a solver to solve to obtain a target effective scene value meeting the confidence coefficient requirement
Figure BDA0003601142830000201
Step 6 solution of deterministic model
The target scene value obtained in the step 5 can satisfy the condition that the joint opportunity constraint (1-1) is established under the condition of certain confidence coefficient, so that the original uncertain model is converted into a deterministic model, and the obtained target scene value is utilized
Figure BDA0003601142830000202
The transformation opportunity constraint obtained is shown as the formula (6-1).
Figure BDA0003601142830000203
Through the steps, the solving model with greatly improved solving efficiency is obtained. And finally, solving a model corresponding to the deterministic constraint by using a mature solver to obtain the optimal value of each decision variable.
In order to make the purpose and technical solution of the present invention more clear, the present invention is further described in detail below with reference to tables, data and simulation cases. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not limiting of the invention.
The effectiveness of the invention is verified by specific simulation cases, and the simulation schemes of the invention are all carried out based on Matlab. The unit combination scheduling problem of the multi-region wind power system considering the correlation of renewable energy sources is taken as an example in the simulation, but the method is also suitable for joint opportunity constraint of other types of problems. The unit combination scheduling problem takes the minimum running cost of the thermal power unit as an objective function, inequality constraints comprise output constraints, climbing constraints, minimum startup and shutdown time constraints and the like, and positive rotation standby constraints of a region are joint opportunity constraints.
The attached table 1 shows the specific parameters of two specific simulation schemes of the present invention. The specific simulation scheme is shown in the attached table 1, and for each scheme, the invention sets the comparison of the solving time and the solving accuracy by adopting the traditional joint opportunity constraint solving method and the method adopted by the invention.
Attached Table 1
Figure BDA0003601142830000204
Figure BDA0003601142830000211
The number of the wind power output historical samples of the two simulation schemes is set to 730, and the minimum rotation standby rate required by the system is 10%. The unit parameters of the two schemes are as follows:
(1) scheme 1: the scheme is that 3 regional power systems of 33 thermal power generating units, every region all contain 1 wind-powered electricity generation field, and 3 regions are connected together through the tie line. Regional 1 electric power system contains 10 thermal power generating units, wind-powered electricity generation installed capacity 850MW, and regional 2 electric power system contains 10 thermal power generating units, and wind-powered electricity generation installed capacity 1050MW, regional 3 electric power system contain 13 thermal power generating units, and wind-powered electricity generation installed capacity 1350MW, the scheduling cycle is 1 hour, and the confidence level that sets up is 90%.
(2) Scheme 2: to compare the effect of different confidence levels on the results, the confidence level was adjusted to 95% based on protocol 1.
Step 1 is implemented: establishing joint opportunity constraints containing multidimensional random variables
Implementation of step 1.1 establishing Positive rotation Standby Joint opportunity constraints
The positive rotation reserve constraint of the multi-region power system requires that the positive rotation reserve of a region k at a time t should be greater than or equal to the positive rotation reserve capacity requirement, and the positive rotation reserve joint opportunity constraint of the region k at the time t is represented as:
Figure BDA0003601142830000212
wherein,
Figure BDA0003601142830000213
and
Figure BDA0003601142830000214
output and rotation reserve of the unit G in the zone k at the moment t, G K The method comprises the following steps of (1) being a set of thermal power generating units;
Figure BDA0003601142830000215
the actual wind power output of the region k at the moment t is a random variable;
Figure BDA0003601142830000216
load power for region k at time t; eta is the rotation reserve rate, the above formula is required to be satisfied under a certain confidence coefficient due to the uncertainty of the random variable, alpha is the confidence coefficient level of the joint opportunity constraint, Pr {. is used for representing the satisfied probability of the constraint event, and the formula (7-1) is required to be Pr {. isThe probability that the positive rotation standby constraints (i.e., K inequalities) of the inner K regions are simultaneously satisfied is greater than or equal to α.
Implementing step 1.2 to convert Joint opportunity constraints into a Standard form
Random variables in the formula (7-1)
Figure BDA0003601142830000221
Move term to the right of the internal inequality, will
Figure BDA0003601142830000222
Moving the term to the left of the internal inequality, we can get:
Figure BDA0003601142830000223
step 2 is implemented: establishing scene set based on random variable historical data
And establishing a multi-dimensional wind power output scene set according to historical wind power data of the multi-region wind power plant, and if each region of the model has one wind power plant, the joint opportunity constraint requires that the positive rotation standby of the n regions simultaneously meet the confidence coefficient requirement, namely the wind power output scene at each moment is n-dimensional.
Step 3 is implemented: and calculating a cumulative distribution function of the random variables by using the historical data samples of the random variables.
Calculating the cumulative distribution function of the wind power random variable of each region according to the step 3, and recording the cumulative distribution function as
Figure BDA0003601142830000224
And (4) implementing the step: and generating a new bundle scene.
And (3) processing the scene generated in the step (2) by adopting a bundle method to obtain a bundle scene set with greatly reduced computational complexity.
And 5, implementation step: and establishing a target scene solving model.
Establishing a target scene solving model by using the bundle scene set obtained in the step 4
And 6, implementation step: and (5) solving a deterministic model.
Using the object scene obtained in step 5
Figure BDA0003601142830000225
And (4) converting the joint opportunity constraint into the deterministic constraint which is easy to solve as shown in the formula (7-3), thereby converting the joint opportunity constraint into the deterministic constraint.
Figure BDA0003601142830000226
And finally, solving the deterministic constraint by using a mature solver to obtain a scheduling result. The advantages of the present invention are further detailed below by comparing the differences between the conventional method and the method of the present invention for processing the two protocols.
The attached table 1 shows the specific contents of the two schemes, and the joint opportunity constraints adopting two different confidence levels of 90% and 95% are solved by respectively adopting the traditional method and the method adopted by the invention, so that the influence of the performance and the confidence difference of the method on the result is verified.
The accompanying tables 2 and 3 show the solving results of the two methods, respectively.
The attached table 2 shows the solution time and the scheduling result cost of the target scene in the first solution scheme of the traditional method and the method of the invention. The attached table 2 shows the solution time for solving the target scene value and the final scheduled cost result when the confidence is 90%. The time spent in solving by adopting the traditional method is 60.05s, while the time spent in solving by adopting the method of the invention is only 31.38s, the solving time is reduced by 47 percent compared with the original time, and for the final scheduling solving result, the cost of the method of the invention is slightly less than that of the traditional method.
Attached table 2
Figure BDA0003601142830000231
The attached table 3 shows the target scene solving time and the scheduling result cost of the second solving scheme by respectively adopting the traditional method and the method of the invention. The attached table 3 shows the solution time for solving the target scene value and the final scheduled cost result when the confidence is 95%. Compared with the traditional method, the method provided by the invention has the advantages that the solving efficiency is greatly improved, and the final scheduling result cost is basically consistent as the case when the confidence coefficient is 90%.
Attached table 3
Figure BDA0003601142830000232
The method improves the solving efficiency on the premise of not losing the solving effectiveness, and realizes the coordination of the solving rapidity and the solving reliability of the scheduling model.
It is noted that when the confidence is 95%, the solving time of both the traditional method and the method of the present invention is less than the solving time when the confidence is 90%, and the improvement of the solving efficiency by using the method of the present invention is also greater than the case when the confidence is 90%, because the uncertainty of the model is reduced along with the improvement of the set joint opportunity constraint confidence, the solving time is reduced; and because the scene number of the generated bundle scene set is also improved along with the improvement of the confidence, the model calculation complexity after the bundle method processing is also greatly reduced.
From the calculation results, the scheduling strategy obtained by the method can improve the solving efficiency without losing the solving accuracy, and the method is proved to be correct and effective.
Fig. 2 is a system architecture diagram for solving a scheduling model of a power system including renewable energy according to an embodiment of the present invention, as shown in fig. 2, including:
a joint opportunity constraint model establishing unit 210, configured to establish a joint opportunity constraint of the power system by using the output of the renewable energy as a random variable; the power system comprises a plurality of renewable energy sources, the random variable is multidimensional, and each dimension of data corresponds to the output of one renewable energy source; the joint opportunity constraint is a constraint relation in a multi-region power system scheduling model, and the power system scheduling model determines scheduling schemes of thermal power generating units and tie lines in each time period by taking the minimum total operating cost of a power system as a target;
a random variable scene set determining unit 220, configured to determine historical data of a corresponding random variable according to historical data of renewable energy output; the renewable energy output corresponding to one historical data of the random variable is used as a scene, the probability of each scene is determined according to the repeated condition of each scene in all the scenes, and the set of different scenes is used as the scene set of the random variable;
an accumulative distribution function determining unit 230, configured to use a historical data of the random variable as a sample, and determine an accumulative distribution function of the random variable by using adaptive kernel density estimation based on the sample of the random variable;
a confidence quantile determining unit 240, configured to determine an inverse function of the corresponding cumulative distribution function based on the cumulative distribution function of the renewable energy output, and substitute a single-opportunity constraint confidence into the inverse function to determine a confidence quantile; the single opportunity constraint confidence is determined from a risk level of the joint opportunity constraint;
a bundle scene set determining unit 250, configured to divide, based on the confidence quantile, an approximate scene that does not affect a final result of the mixed integer programming solution in the scene set from an effective scene to obtain a scene set in which each dimensional random variable is bundle-processed; synthesizing the approximate scenes in the processed scene set into a bundle, so that the bundle and the effective scenes in the processed scene set jointly form a bundle scene set of random variables;
the effective scene set determining unit 260 is configured to solve the minimum sum of the scene values of the dimensions of the bundle scene set generated by the renewable energy output to obtain a target effective scene value meeting the confidence requirement;
and a deterministic model solving unit 270, configured to convert the joint opportunity constraint of the power system into a deterministic constraint by using the target effective scene value, and solve a power system scheduling model corresponding to the deterministic constraint.
It is understood that detailed functional implementation of each unit described above can refer to the description in the foregoing method embodiment, and is not described herein again.
It will be understood by those skilled in the art that the foregoing is only a preferred embodiment of the present invention, and is not intended to limit the invention, and that any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the present invention.

Claims (10)

1. A method for solving a power system scheduling model containing renewable energy is characterized by comprising the following steps:
the output of renewable energy sources is taken as a random variable, and joint opportunity constraint of the power system is established; the power system comprises a plurality of renewable energy sources, the random variable is multidimensional, and each dimension of data corresponds to the output of one renewable energy source; the joint opportunity constraint is a constraint relation in a multi-region power system scheduling model, and the power system scheduling model determines scheduling schemes of thermal power generating units and tie lines in each time period by taking the minimum total operating cost of a power system as a target;
determining historical data of corresponding random variables according to historical data of renewable energy output; the renewable energy output corresponding to one historical data of the random variable is used as a scene, the probability of each scene is determined according to the repeated condition of each scene in all the scenes, and the set of different scenes is used as the scene set of the random variable;
taking historical data of the random variable as a sample, and determining the cumulative distribution function of the random variable by adopting self-adaptive kernel density estimation based on the sample of the random variable;
determining an inverse function of the corresponding accumulative distribution function based on the accumulative distribution function of the renewable energy output, and substituting the single-opportunity constraint confidence coefficient into the inverse function to determine a confidence coefficient quantile point; the single opportunity constraint confidence is determined from a risk level of the joint opportunity constraint;
based on the confidence quantile, dividing an approximate scene which does not influence the final result of the mixed integer programming solution in the scene set and an effective scene to obtain the scene set of which each dimension of random variables is subjected to bundle processing; synthesizing the approximate scenes in the processed scene set into a bundle, so that the bundle and the effective scenes in the processed scene set jointly form a bundle scene set of random variables;
solving to obtain a target effective scene value meeting the confidence coefficient requirement by taking the minimum sum of all dimensional scene values of a bundle scene set generated by the output of renewable energy as a target function;
and converting the joint opportunity constraint of the power system into a deterministic constraint by using the target effective scene value, and solving a power system scheduling model corresponding to the deterministic constraint.
2. The method according to claim 1, wherein the joint opportunity constraint of the power system is established by taking the output of the renewable energy source as a random variable, specifically:
the general expression of the joint opportunity constraint is:
Figure FDA0003601142820000021
P G a vector representing the output of the thermal power generating unit, y represents a vector corresponding to a parameter other than the renewable resource output and the thermal power generating unit output, P R Is an uncertain renewable energy output variable, which is a K-dimensional random variable, g k (P G Y) is a constraint equation containing the output of the thermal power generating unit, g k (P G ，y)≥-P R To constrain events, (1- α) JCC ) Is the joint opportunity constraint confidence, α JCC Is the risk level of the joint opportunity constraint, Pr {. represents the probability of satisfaction of the constraint event, and the joint opportunity constraint requires that the probability of satisfaction of K inequalities inside Pr {. is greater than or equal to (1-alpha) JCC )。
3. The method of claim 1, wherein the scene set of random variables is determined by:
acquiring or updating the historical data of the output of the renewable energy sources, and acquiring a scene set capable of representing the randomness of the output of the renewable energy sources according to the historical data; let K dimension random variable history data
Figure FDA0003601142820000022
The resulting limited set of scenes is C,
Figure FDA0003601142820000023
the method is a scene in the historical data of the output of the renewable energy, wherein si belongs to C,
Figure FDA0003601142820000024
performing statistical clustering on the same scene in the scene set for the output scene value of the kth-dimension renewable energy in the sih scene to obtain a new scene set S, wherein S belongs to S,
Figure FDA0003601142820000025
a scenario in the historical data of new renewable energy output, wherein
Figure FDA0003601142820000026
The output scene value of the kth dimension renewable energy source of the S scene in the new scene set S, wherein the S scene comprises an initial scene d si Is M s ；
If the total number of scenes in C is Sn, then scene d in S s Probability P of s Comprises the following steps: p s ＝M s /Sn s∈S；
Figure FDA0003601142820000027
And acquiring a scene set S generated by the renewable energy output historical data and the corresponding probability of each scene in the scene set S.
4. The method according to claim 1, wherein the stochastic variable-based samples are adapted to determine a cumulative distribution function of the stochastic variables using adaptive kernel density estimation, in particular:
calculating the fixed bandwidth of the adaptive kernel density estimation by adopting the principle of minimum mean integral square error;
calculating cumulative distribution function of renewable energy output based on the fixed bandwidth
Figure FDA0003601142820000031
Figure FDA0003601142820000032
Figure FDA0003601142820000033
Wherein, N is the number of samples,
Figure FDA0003601142820000034
is a kernel function, H is a fixed bandwidth under the kernel function,
Figure FDA0003601142820000035
cumulative distribution function for fixed bandwidth, x is an independent variable representing renewable energy output, x i The ith sample value of the renewable energy output is obtained;
dividing N samples into M intervals, continuously iterating the intervals in an increasing mode from small to large, and determining the number of the intervals by judging whether the fitting error is smaller than an error threshold value;
calculating a position parameter h of each interval j ：
Figure FDA0003601142820000036
Wherein, the j, l interval is numbered,
Figure FDA0003601142820000037
an abscissa corresponding to the median of the sample cumulative distribution of the interval l; the sample bandwidth of each interval becomes H × H j Beta is a sensitivity coefficient, and beta is more than 0 and less than 1;
self-adaptive bandwidth kernel density estimation is carried out by utilizing position parameters of each interval, and renewable energy sources are calculatedProbability density function of force
Figure FDA0003601142820000038
Figure FDA0003601142820000039
Wherein N is j I is the number of samples in the interval j, i is the number of sample points in the interval j,
Figure FDA00036011428200000310
computing probability density functions for kernel estimation
Figure FDA00036011428200000311
With the true distribution f (x) i ) Mean square error MSE of (a):
Figure FDA00036011428200000312
the larger the partition number M is, the finer the bandwidth correction is, and the lower the error of the overall fitting is; by setting the error threshold MSE T When MSE > MSE T Then, the number M of the intervals is continuously increased, the step length of the intervals is increased, and the position parameters and the probability density function are repeatedly calculated until MSE is less than MSE T Until or reaching the maximum number of partitions, the maximum number of partitions is the number of samples, and the number of intervals obtained at the moment is used as the number of intervals which are determined finally;
after the interval number is finally determined, the renewable energy output P is calculated by utilizing interval self-adaptive bandwidth kernel density estimation R The cumulative distribution function of (1) is written as
Figure FDA00036011428200000313
Figure FDA00036011428200000314
x k And outputting the k-dimension renewable energy.
5. The method according to claim 1, wherein the bundle scene set of the random variables is determined by:
calculating the inverse function of the cumulative distribution function of the renewable energy output of each dimension, and recording as
Figure FDA0003601142820000041
Constraining each single opportunity to confidence 1-alpha ICC Substituting into the inverse function of the cumulative distribution function to obtain confidence quantile points
Figure FDA0003601142820000042
Figure FDA0003601142820000043
Quantile based on confidence
Figure FDA0003601142820000044
Dividing an approximate scene which does not influence the final result of the mixed integer programming solution in the original scene set and an effective scene, and respectively processing each dimension random variable, wherein the processing method comprises the following steps:
Figure FDA0003601142820000045
thus, s scenes of random variables of each dimension after bundle preprocessing are obtained;
after the scenes of the random variables of each dimension are divided, a plurality of approximate scenes before the confidence quantile and an effective scene set after the confidence quantile are formed, and the approximate scenes are gathered into a new bundle, wherein the method comprises the following steps:
Figure FDA0003601142820000046
wherein s is b Representing bundle of renewable energy output scenes, wherein Sn is the total number of scenes of the original scene set, s e Representing the effective scene set of renewable energy output, and the two jointly form a new bundle scene set s B Each scene having a probability of
Figure FDA0003601142820000047
Wherein,
Figure FDA0003601142820000048
representing the ith scene of the bundle scene set, wherein each scene comprises k-dimensional renewable energy output scene values; bn represents the number of scenes constituting a bundle; bn represents the number of scenes that will newly generate the bundle scene set, and Sn is Bn + Bn-1.
6. The method according to claim 5, wherein the solving obtains a target effective scene set that satisfies the confidence requirement, specifically:
the objective function is set up as follows:
Figure FDA0003601142820000051
the objective function requires that the sum of the dimensions of the scene generated by the renewable energy output is minimal,
Figure FDA0003601142820000052
is the k-dimension target scene value;
Figure FDA0003601142820000053
Figure FDA0003601142820000054
representing a kth dimension scene value in the ith scene of the Bundle scene set, and requiring that a target scene value is larger than any Bundle scene value meeting the condition;
Figure FDA0003601142820000055
the sum of the corresponding probabilities of all scenes in the scene set is required to be 1;
Figure FDA0003601142820000056
Figure FDA0003601142820000057
is a binary variable, and
Figure FDA0003601142820000058
with historical scenes
Figure FDA0003601142820000059
Correspondingly, each history scene
Figure FDA00036011428200000510
All have corresponding thereto
Figure FDA00036011428200000511
Figure FDA00036011428200000512
Requiring the ith target value to be greater than the product of the binary variable and the historical scene;
substituting the objective function into a solver to solve based on the constraint condition to obtain an effective objective scene value meeting the confidence requirement
Figure FDA00036011428200000513
7. The method according to claim 6, wherein the solving of the power system model is specifically:
the target effective scene value can satisfy the condition that the joint opportunity constraint is established under a certain confidence coefficient condition, so that an uncertain model corresponding to the original joint opportunity constraint is converted into a deterministic model, wherein the obtained target scene value is utilized
Figure FDA00036011428200000514
The resulting transformation opportunity constraints are:
Figure FDA00036011428200000515
and solving the deterministic model by using a solver to obtain the optimal value of each decision variable.
8. A power system scheduling model solving system containing renewable energy sources is characterized by comprising:
the joint opportunity constraint model establishing unit is used for establishing joint opportunity constraint of the power system by taking the output of the renewable energy as a random variable; the power system comprises a plurality of renewable energy sources, the random variable is multidimensional, and each dimension of data corresponds to the output of one renewable energy source; the joint opportunity constraint is a constraint relation in a multi-region power system scheduling model, and the power system scheduling model determines scheduling schemes of thermal power generating units and tie lines in each time period by taking the minimum total operating cost of a power system as a target;
the random variable scene set determining unit is used for determining historical data of corresponding random variables according to historical data of renewable energy output; the renewable energy output corresponding to one historical data of the random variable is used as a scene, the probability of each scene is determined according to the repeated condition of each scene in all the scenes, and the set of different scenes is used as the scene set of the random variable;
the cumulative distribution function determining unit is used for taking historical data of the random variable as a sample and determining the cumulative distribution function of the random variable by adopting self-adaptive kernel density estimation based on the sample of the random variable;
the confidence quantile determining unit is used for determining an inverse function of the corresponding accumulative distribution function based on the accumulative distribution function of the renewable energy output and substituting the single-opportunity constraint confidence into the inverse function to determine a confidence quantile; the single opportunity constraint confidence is determined from a risk level of the joint opportunity constraint;
the bundle scene set determining unit is used for dividing an approximate scene which does not influence the final result of the mixed integer programming solving in the scene set and an effective scene based on the confidence quantile points to obtain a scene set of each dimension of random variables after bundle processing; synthesizing the approximate scenes in the processed scene set into a bundle, so that the bundle and the effective scenes in the processed scene set jointly form a bundle scene set of random variables;
the effective scene set determining unit is used for solving to obtain a target effective scene value meeting the confidence coefficient requirement by taking the minimum sum of all dimensional scene values of the bundle scene set generated by the renewable energy output as a target function;
and the deterministic model solving unit is used for converting the joint opportunity constraint of the power system into a deterministic constraint by using the target effective scene value and solving the power system scheduling model corresponding to the deterministic constraint.
9. The system according to claim 8, wherein the bundle scene set determining unit determines the bundle scene set of the random variables by: calculating the inverse function of the cumulative distribution function of the renewable energy output of each dimension, and recording as
Figure FDA0003601142820000061
Constraining each single opportunity to confidence 1-alpha ICC Substituting into the inverse function of the cumulative distribution function to obtain confidence quantile points
Figure FDA0003601142820000071
Figure FDA0003601142820000072
Figure FDA0003601142820000073
Quantile based on confidence
Figure FDA0003601142820000074
Dividing an approximate scene which does not influence the final result of the mixed integer programming solution in the original scene set and an effective scene, and respectively processing each dimension random variable, wherein the processing method comprises the following steps:
Figure FDA0003601142820000075
thus, s scenes of random variables of each dimension after bundle preprocessing are obtained; after the scenes of the random variables of each dimension are divided, a plurality of approximate scenes before the confidence quantile and an effective scene set after the confidence quantile are formed, and the approximate scenes are gathered into a new bundle, wherein the method comprises the following steps:
Figure FDA0003601142820000076
wherein s is b Representing bundle of renewable energy output scenes, wherein Sn is the total number of scenes of the original scene set, s e Representing the effective scene set of renewable energy output, and the two jointly form a new bundle scene set s B Each scene having a probability of
Figure FDA0003601142820000077
Wherein,
Figure FDA0003601142820000078
representing an ith scene of the bundle scene set, wherein each scene comprises k-dimensional renewable energy output scene values; bn represents the number of scenes constituting a bundle; bn denotes the number of scenes newly generating the bundle scene set, and Sn is Bn + Bn-1.
10. The system according to claim 9, wherein the effective scene set determining unit solves the target effective scene set satisfying the confidence requirement, specifically: the objective function is set up as follows:
Figure FDA0003601142820000079
the objective function requires that the sum of the dimensions of the scene generated by the renewable energy output is minimal,
Figure FDA00036011428200000710
is the k-dimension target scene value;
Figure FDA00036011428200000711
Figure FDA00036011428200000712
representing a kth dimension scene value in the ith scene of the Bundle scene set, and requiring that a target scene value is larger than any Bundle scene value meeting the condition;
Figure FDA0003601142820000081
the sum of the corresponding probabilities of all scenes in the scene set is required to be 1;
Figure FDA0003601142820000082
Figure FDA0003601142820000083
is a binary variable, and
Figure FDA0003601142820000084
with historical scenes
Figure FDA0003601142820000085
Correspondingly, each history scene
Figure FDA0003601142820000086
All have delta corresponding to it i s ；
Figure FDA0003601142820000087
Requiring the ith target value to be greater than the product of the binary variable and the historical scene; substituting the target function into a solver to solve based on the constraint condition to obtain a target effective scene value meeting the confidence coefficient requirement
Figure FDA0003601142820000088
CN202210403184.0A 2022-04-18 2022-04-18 A method and system for solving a power system dispatching model containing renewable energy Active CN114819319B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210403184.0A CN114819319B (en) 2022-04-18 2022-04-18 A method and system for solving a power system dispatching model containing renewable energy

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210403184.0A CN114819319B (en) 2022-04-18 2022-04-18 A method and system for solving a power system dispatching model containing renewable energy

Publications (2)

Publication Number Publication Date
CN114819319A true CN114819319A (en) 2022-07-29
CN114819319B CN114819319B (en) 2024-08-20

Family

ID=82536585

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210403184.0A Active CN114819319B (en) 2022-04-18 2022-04-18 A method and system for solving a power system dispatching model containing renewable energy

Country Status (1)

Country Link
CN (1) CN114819319B (en)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20170308831A1 (en) * 2016-04-20 2017-10-26 Aspen Technology, Inc. Robust feedstock selection system for the chemical process industries under market and operational uncertainty
CN112865170A (en) * 2020-12-31 2021-05-28 南京理工大学 Scene probability-based load recovery optimization method considering wind power output correlation
CN114069692A (en) * 2021-10-18 2022-02-18 广西电网有限责任公司 Joint opportunity constraint optimization method and device for solving power scheduling problem

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20170308831A1 (en) * 2016-04-20 2017-10-26 Aspen Technology, Inc. Robust feedstock selection system for the chemical process industries under market and operational uncertainty
CN112865170A (en) * 2020-12-31 2021-05-28 南京理工大学 Scene probability-based load recovery optimization method considering wind power output correlation
CN114069692A (en) * 2021-10-18 2022-02-18 广西电网有限责任公司 Joint opportunity constraint optimization method and device for solving power scheduling problem

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
MIGUEL LEJEUNE: "Mathematical programming approaches for generating p-efficient points", 《EUROPEAN JOURNAL OF OPERATIONAL RESEARCH》, vol. 207, no. 2, 13 January 2011 (2011-01-13), pages 590 - 600, XP027169486 *
YAN SUN: "An Improved Chance-Constrained Method for Unit Commitment in Multi-regional Power Systems Considering Wind Power Uncertainties", 《2022 5TH INTERNATIONAL CONFERENCE ON RENEWABLE ENERGY AND POWER ENGINEERING》, 21 November 2022 (2022-11-21), pages 1 - 7 *

Also Published As

Publication number Publication date
CN114819319B (en) 2024-08-20

Similar Documents

Publication Publication Date Title
CN113193600B (en) Electric power system scheduling method and device considering wind power probability distribution uncertainty
CN113344283B (en) Evaluation method for new energy consumption capacity of Energy Internet based on edge intelligence
CN119297976B (en) A multi-stage unit combination method for power systems containing a high proportion of renewable energy
CN119298194A (en) A coordinated control method and system for island microgrid with multi-type energy complementation
Su et al. Short-term transmission capacity prediction of hybrid renewable energy systems considering dynamic line rating based on data-driven model
CN118920603A (en) Two-stage robust optimal scheduling method for source load power multi-interval uncertainty set
CN111585279A (en) Microgrid optimization scheduling method based on new energy consumption
CN109299862B (en) A kind of convex loose appraisal procedure of wind-powered electricity generation maximum digestion capability
CN117394448A (en) Power system dispatching method and medium based on robust security domain optimization
CN119813386B (en) Multi-objective real-time operation optimization scheduling method and system for microgrids
CN115051388A (en) Distribution robustness-based &#39;source-network-load-storage&#39; two-stage scheduling optimization method
CN120198150A (en) A method, device, equipment and storage medium for assisting decision-making management of power trading
CN109255102B (en) Photovoltaic power generation maximum grid-connected capacity evaluation method based on convex relaxation
Lin et al. Scenarios analysis method based on wind and photovoltaic power output correlation
CN114819319A (en) Electric power system scheduling model solving method and system containing renewable energy
Xu et al. Flexibility balance mechanism and flexibility requirement quantification of power system with high share of uncertainty power
Chen et al. Power grid dispatch with high proportion of new energy integration
CN114977318B (en) Random-deterministic coupled power supply scheduling method and system considering correlation
Gou et al. Research on optimal dispatching method of power systems with high proportion of clean energy based on multi-energy complementarity
Qiu et al. Fuzzy Logic based scheduling optimization and risk control in power systems
Sun et al. Multi-time scale stochastic optimization for hybrid AC-DC distribution networks with PET based on EC-Kmeans clustering
CN117833300B (en) A method for configuring energy storage capacity based on intelligent power distribution equipment
Hu et al. Multi-Time-Scale Dispatching Method for Power System Source-Load Considering Wind Power Correlation
Han et al. Intraday rolling optimal dispatch model of power system considering supply-demand interaction
CN110874611A (en) Grading compensation method for ultra-short-term forecast error level of wind power cluster power

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant