CN114818205B - An online sensing method for tip clearance in the whole life cycle of aero-engine - Google Patents
An online sensing method for tip clearance in the whole life cycle of aero-engine Download PDFInfo
- Publication number
- CN114818205B CN114818205B CN202210732109.9A CN202210732109A CN114818205B CN 114818205 B CN114818205 B CN 114818205B CN 202210732109 A CN202210732109 A CN 202210732109A CN 114818205 B CN114818205 B CN 114818205B
- Authority
- CN
- China
- Prior art keywords
- model
- tip clearance
- engine
- turbine
- blade tip
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/17—Mechanical parametric or variational design
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/23—Design optimisation, verification or simulation using finite element methods [FEM] or finite difference methods [FDM]
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/04—Ageing analysis or optimisation against ageing
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/08—Thermal analysis or thermal optimisation
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Geometry (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- Computer Hardware Design (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Turbine Rotor Nozzle Sealing (AREA)
Abstract
本发明涉及航空发动机稳定性控制技术领域，具体为一种航空发动机全生命周期叶尖间隙在线感知方法，本发明通过发展一种航空发动机机载工作参数感知机理模型部件特性自修正方法，获得极限环境下高精度航空发动机工作参数机载实时感知基线模型；再通过合理假设和简化叶尖间隙的建模过程，分析部件变形机理且考虑材料特性等因素，建立高置信度的涡轮叶尖间隙模型；最后设计基于涡轮健康估计器实时跟踪发动机的退化情况，融合深度卷积网络学习算法建立叶尖间隙估计器修正叶片、轮盘和机匣的蠕变情况，最终形成一种航空发动机全生命周期叶尖间隙在线感知方法。
The invention relates to the technical field of aero-engine stability control, in particular to an on-line sensing method for blade tip clearance in the whole life cycle of an aero-engine. High-precision airborne real-time perception baseline model of aero-engine operating parameters in the environment; then through reasonable assumptions and simplifying the modeling process of tip clearance, analyzing the deformation mechanism of components and considering factors such as material properties, a high-confidence turbine tip clearance model is established ;The final design is based on the turbine health estimator to track the degradation of the engine in real time, and the deep convolutional network learning algorithm is integrated to establish a tip clearance estimator to correct the creep of the blades, discs and casings, and finally form an aero-engine full life cycle. On-line sensing method of tip clearance.
Description
Technical Field
The invention relates to the technical field of stability control of an aero-engine, in particular to an on-line sensing method for tip clearance of the aero-engine in a full life cycle.
Background
The aircraft engine is the heart of an airplane, the development level of the aircraft engine is one of the centralized embodiments of national comprehensive strength, industrial foundation and technology level, the aircraft engine is an important strategic guarantee of national safety and strong national status, and the aircraft engine has important significance for the development of scientific technology and national economy.
Conventional engine control is based on sensor control, i.e. engine operation is indirectly controlled by measurable engine state parameters such as speed, pressure ratio and temperature. Although the control mode is simple and reliable, the variation condition of the unmeasured parameters, particularly the blade tip clearance, in the working process of the engine is difficult to accurately reflect, the margin loss of the engine in the worst working environment must be considered in the design, and the full exertion of the performance of the engine is limited. With the rapid development of a Full Authority Digital Electronic Control (FADEC) technology, the performance of the traditional engine control is brought into full play, and in order to further improve the overall performance of the engine from the control perspective, the difficultly measured parameter of the blade tip clearance needs to be sensed in real time and actively controlled in a closed loop manner, so that the potential of the engine is fully exploited.
However, at present, most of the research on model-based blade tip clearance sensing at home and abroad focuses on the simulation stage, and is rarely reported in the practical application of engineering. The reason mainly includes the following two aspects: firstly, the accuracy of an airborne model of an aeroengine is insufficient, so that errors exist between the acquired part characteristics and a real engine, the estimation result of the airborne model is inaccurate, and the input accuracy of a blade tip clearance model is low; secondly, the high-confidence real-time perception of the blade tip clearance of the aero-engine with the full life cycle has many defects, such as dynamic uncertainty of the model, modeling error of an airborne real-time model, individual difference between engines of the same model caused by manufacturing and installation tolerance of the engine and the like, so that the prediction precision of the blade tip clearance model is not high, and the degradation phenomenon of part characteristic change of a real engine due to long-time operation cannot be reflected in a traditional model.
Disclosure of Invention
The method is just directed at the non-contact type prediction method of the blade tip clearance, the degradation problem is mostly not considered, namely the blade tip clearance is only considered to be estimated according to the blade tip clearance change condition of an aeroengine under a single working condition at a certain moment, and the method has little significance to practical application; in addition, in the tip clearance estimation method for predicting and considering the degradation condition published at present, taking 'turbine tip clearance estimation method research considering the performance degradation of an engine' as an example, the used method is always an empirical formula, namely, the temperature change condition is judged through the running time of the engine, so that the change considering the temperature is put into the tip clearance calculation to obtain the tip clearance value considering the degradation. However, the airborne problem is not solved, and the main reasons are: 1) the temperature change conditions caused by the same operation duration of different aero-engines are different inevitably, and even the same type of aero-engine experiences the same flight time, because the working environment cannot be completely the same, the high accuracy of the estimation method cannot be ensured at all by judging through the simple empirical rule of blade tip clearance = f (operation duration); 2) the complex calculation can bring accurate calculation results but cannot meet the real-time requirement required in an onboard computer, so that the onboard application of blade tip clearance estimation cannot be realized by using a blade tip clearance estimation method on line. Therefore, the method aims to realize the blade tip clearance estimation of the airborne aeroengine and solve the online estimation of all blade tip clearances during the operation of the engine.
In order to solve the problems, the invention provides an online sensing method for blade tip clearances in a full life cycle of an aircraft engine, which is characterized by comprising the following specific steps:
s1, establishing an airborne real-time perception baseline model of the working parameters of the aero-engine;
s11, self-correction of the characteristics of the airborne model component is realized by evaluating the performance of the test data of the whole engine bench test;
s12, establishing a bench test database for evaluation to obtain an aircraft engine airborne baseline model with high confidence level, and simultaneously carrying out linearization on the aircraft engine airborne baseline model to construct a discrete linear time-invariant aircraft engine state space equation, wherein the method comprises the following steps:
x
k+1
=Ax
k
+Bu
k
+Lh
k
+w
k
y
k
=Cx
k
+Du
k
+Mh
k
+v
k；
wherein the subscriptkRepresenting a discrete-state aircraft engine modelkThe data acquisition at the time of the next time,xis that the state variable vector of the aircraft engine comprises the rotating speed of a low-pressure rotorN f And high pressure rotor speedN c ，uIs that the control quantity, i.e. the input deviation, includes the fuel quantityW F Area of the nozzle tipA 8 ；yThe measured output quantity, namely the actual model output comprises high-low pressure rotating speedN c 、N f (ii) a Total temperature and pressure at fan outletT 22 、P 22 Total temperature and total pressure at outlet of gas compressorT 3 、P 3 Total pressure at outlet of combustion chamberP 4 And total pressure of total temperature at outlet of high-pressure turbineT 45 、P 45 Vector of motionhRepresenting fan efficiency and flow as health parameters of an aircraft engine requiring estimator estimationη f 、W f (ii) a Compressor efficiency and flow
η c 、W c High pressure turbine efficiency and flowη t 、W t ；w k 、v k Respectively a noise matrix and a measurement noise matrix of the system; the matrix A, B, C, D, L, M is the engine linear model coefficients; the health parameter represented by the vector h is an unknown input to the system;
s2, constructing a thermodynamic and rotor dynamics equation, and calculating radial deformation of a casing, a wheel disc and blades to obtain a mathematical model representing the change of the tip clearance of the turbine in real time, namely an aeroengine tip clearance sensing model;
s3, building a turbine health parameter estimator and a blade tip clearance degradation estimator; supplementing the output of the tip clearance reference model through a neural network and finally establishing an aero-engine full life cycle tip clearance online perception model based on model prediction;
the baseline model and the turbine health estimator are combined in the form:
preferably, S3 further includes the following steps:
s301, starting a turbine health estimator, and constructing all health parametershLinear combination of (2), i.e. model adjustment parameter vectorq=V*hWhereinV * Is a transformation matrix for constructing the adjustment parameter vector, the health parameter vectorh=V *-1 qTherefore, it is only necessary to obtainV * An estimated value of the health parameter can be obtained;
s302, constructing an aeroengine model to adapt to the estimation of the health parameters of the turbine, namely changing the model into the following form:
s303, constructing an estimator model, and obtaining the predicted output of the estimator through the estimator model, wherein
P ∞ =A xq P ∞ A xq T -A xq P ∞ C xq T （C xq P ∞ C xq T +R） -1 C xq P ∞ C xq T +Q xq ，
K
∞
=P
∞
C
xq
T
（C
xq
P
∞
C
xq
T
+R）
-1
；
S304, evaluation according to S303K ∞ 、P ∞ As a result of the calculation of (a) is,
then the sum of the squared errors can be calculated as
；
S305, in each iteration, evaluating the change of the sum of squared errors relative to the previous iteration to determine whether the output error achieves convergence within a specified range, if so, skipping S306 and continuing to execute S307, otherwise, updating S306V * ；
S306, updating through a nonlinear least square functionV * Require
=1 and return to S301;
s307, returning after convergenceV * And ending the flow in the turbine health estimator;
s308, output by the turbine health estimatorV * Substituting into formulah=V *-1 qCalculating to obtain an estimate of a turbine health parameterh={η f ，W f ，η c ，W c ，η t ，W t }，Respectively representing the efficiency and flow of a fan, an air compressor and a high-pressure turbine, correcting the health parameters of an onboard baseline model of the engine, and simultaneously obtaining the output of an actual baseline modely＇={N c ＇ ,m ＇ HPC ,T 3 ＇ , m ＇ core , T 4 ＇ }，Respectively representing the high-pressure rotor speed after degradationN c ＇ ，Induced air flow of compressorm ＇ HPC ，Bleed air temperature of compressorT 3 ＇ ，Turbine inlet flowm ＇ core ,And turbine inlet temperatureT 4 ＇ ；
S309, establishing a calculation module by considering the surface factors of the outer wall, the inner wall and the wheel disc of the casing, establishing a blade tip clearance reference model, and outputting the reference model through an actual baseline modely＇={N c ＇ ,m ＇ HPC ,T 3 ＇ , m ＇ core , T 4 ＇ }As input of the tip clearance reference model, the definition is convenient for distinguishing
={N c ＇ ,m ＇ HPC ,T 3 ＇ , m ＇ core , T 4 ＇ }So that
Obtaining tip clearance estimate output for input to a tip clearance model
；
S310, building a deep convolution neural network, setting the learning rate to be 0.01, training the batch to be 2000, iterating for 300 times, setting the output layer as the output parameter tip clearance correction value, adjusting the number of nodes of the hidden layer according to the training effect of the model, setting the initial value to be 8, and inputting the initial value to be the output of the actual baseline modely＇={N c ＇ ,m ＇ HPC ,T 3 ＇ , m ＇ core , T 4 ＇ }Definition ofx＇As inputs to neural network models，x＇={N c ＇ ,m ＇ HPC ,T 3 ＇ , m ＇ core , T 4 ＇ }Compensating the blade tip clearance of the engine nonlinear model through a deep convolution network to obtain a blade tip clearance correction valuer'；
S311, obtaining high-precision input parameters by using tip clearance perception modelr'Calculating according to thermal deformation modules of blades and casings in the built blade tip clearance model sensing module, and adding a blade tip clearance correction value obtained through a neural network model
Finally obtaining accurate online blade tip clearance estimation value
；
S312, a bench test and an evaluation platform are built to verify the accuracy of the parameter perception model in sensing the engine parameters, test run data are stored through a data analysis and data storage module, and after an engine test run experiment is completed once, the accuracy of the parameter perception model of the aero-engine is evaluated in a deviation mode.
Preferably, the airborne real-time perception baseline model of the aircraft engine operating parameters established in S1 includes an airborne real-time perception module, a data perception module and a data management and visualization module.
Preferably, in S2, taking into account the geometrical characteristics of the turbine blade tip clearance key part and the material characteristics changing with the temperature, the thermodynamic and rotor dynamics equation is constructed by a one-dimensional finite difference method and a concentrated mass method.
Compared with the prior art, the invention has the following beneficial technical effects:
(1) and carrying out online performance evaluation through measurable parameters to obtain a high-precision baseline airborne model, and then carrying out accurate estimation on the blade tip clearance in the whole life cycle under the condition of considering degradation through the built state estimator.
(2) Health parameters and section output parameters of the engine are used as input through a deep convolution neural network algorithm, the estimated value of the blade tip clearance is used as output to perform off-line learning to establish a neural network model adaptive to estimation of the blade tip clearance degradation amount, the estimated value of the blade tip clearance of a reference model is corrected on line, and the confidence coefficient and the real-time performance of the engine blade tip clearance perception are improved to meet the requirement of airborne application.
Drawings
FIG. 1 is a model prediction-based full-lifecycle tip clearance sensing architecture for an aircraft engine according to the present invention;
FIG. 2 is a schematic block diagram of a baseline airborne model of an aircraft engine according to the present invention;
FIG. 3 is a flow chart of blade tip clearance modeling of an aircraft engine turbine component according to the present invention.
Detailed Description
Example 1
As shown in fig. 1-3, the aeroengine full-life-cycle tip clearance online sensing method provided by the invention is characterized in that an aeroengine working parameter sensing model mainly comprises three modules: the system comprises an airborne real-time sensing module, a data sensing module and a data management and visualization module. The perception model formed by the three modules can intelligently perceive the parameters of the aircraft engine under the condition of the limit working condition, and the advanced control and the fine design of the engine are realized as the research background. The method aims at the problems that the testable space in the turbine of the engine is limited, the blade tip clearance cannot be sensed, and the change of the blade tip clearance cannot be accurately tracked after the performance of the engine is degraded by the conventional parameter sensing method based on the model.
The full-lifecycle blade tip clearance sensing method of the aero-engine based on model prediction comprises the following steps:
(1) airborne real-time sensing baseline model for working parameters of aero-engine
Considering the problem that the airborne model is not matched with the standard real engine pre-delivery parameters, the invention realizes the self-correction of the characteristics of the airborne model component by performing performance evaluation on the test data of the whole engine bench test, thereby obtaining a high-precision baseline airborne model. FIG. 2 is a schematic block diagram of a reference airborne real-time model of an aircraft engine based on test data under a test run condition of a complete machine bench. First, the method obtains the rotation speed according to a pressure/temperature/thrust/flow sensor arranged on a rack (N f 、N c ) Temperature (c)T 1 、T 13 、T 23 、T 3 、T 16 、T 6 ) Pressure, pressure(P amb 、P 1 、P 13 、P 23 、P 3 、P 16 、P 6 ) Flow rate (b)W 2 、W fb ) Thrust (1)F) Equal test data as input for engine performance evaluation. The main task of test data evaluation is to calculate characteristic parameters of each part, aerodynamic and thermal parameters of each section, critical section area and the like one by one from an engine inlet to an engine outlet according to test measurement data.
(2) Aero-engine blade tip clearance perception model
Based on a first principle, the geometric characteristics of a Turbine blade Tip Clearance key part and the material characteristics changing along with temperature are considered, a thermodynamic and rotor dynamics equation is constructed by using a one-dimensional finite difference method, a concentrated mass method and other methods, a radial deformation of a casing, a wheel disc and a blade is calculated, a mathematical model for representing the change of the Turbine blade Tip Clearance in real time is obtained, a detailed modeling flow of the blade Tip Clearance is shown in FIG. 3, and the reference is Integrated Turbine Tip Clearance and Gas Turbine Engine Simulation.
And respectively calculating the thermodynamic deformation model and the rotor kinetic deformation model of each part by the method, and integrating to finally obtain the turbine blade tip clearance model. The blade tip clearance model takes the high-pressure turbine of the engine as a simulation object, and the calculation result of the nonlinear component-level mathematical model of the engine is taken as the input quantity of the high-pressure turbine of the engine, so that the blade tip clearance simulation calculation of the high-pressure turbine of the engine can be carried out.
(3) Model prediction-based aeroengine full-life blade tip clearance sensing model
The reference model in the content 2 only completes the blade tip clearance estimation value under a single flight condition at a single moment, and a turbine health parameter estimator and a blade tip clearance degradation estimator are required to be built if the blade tip clearance in the full life cycle is estimated at an online high confidence level.
The turbine health parameter estimator can calculate the engine health parameters through the deviation between the performance parameters output by the real engine and the performance parameters output by the engine baseline model, because the engine can generate component loss or degradation in operation, the phenomenon can be reflected on the performance parameters of the real engine, and the baseline model needs to correct the health parameters of the rotor component through the turbine health parameter estimator due to constant value setting, so that the error of the output performance parameters of the baseline model and the real engine model is lower than the maximum limit error which can be accepted by the engine, an airborne model can be provided for high-confidence-level input in the full life cycle range of the aircraft engine of a reference model, and the blade tip clearance estimation can be accurately carried out.
The health parameter estimator estimates the overall performance degradation of the part and provides the estimated performance degradation to the airborne model, if the blade, the wheel disc and the casing run suddenly due to long-time high-intensity operation and creep leads to insufficient input precision of the reference model, the output of the blade tip clearance reference model needs to be supplemented through the neural network, and the blade tip clearance degradation estimator reflects the high-order mapping relation between the degraded performance parameters and the creep of the blade, the casing and the wheel disc of the engine airborne model by building the deep convolutional neural network. And assuming that the aircraft engine creeps after the maximum state or the stress application state, the performance parameters passing through the turbine health estimator can obtain the increment of the blade tip clearance after passing through the deep neural network model, and the increment is supplemented to the output of the blade tip clearance reference model to obtain the estimated blade tip clearance value.
The entire model is composed of the onboard baseline model in fig. 2, the tip clearance reference model in fig. 3, the turbine health estimator, and the tip clearance degradation estimator collectively, and is finally presented in fig. 1.
The specific implementation process is as follows:
1. a bench test database is established in the mode of FIG. 2, an aircraft engine airborne baseline model with high confidence is evaluated and obtained by the method, and meanwhile, the aircraft engine airborne baseline model is linearized to construct a discrete linear time-invariant aircraft engine state space equation as follows:
x
k+1
=Ax
k
+Bu
k
+Lh
k
+w
k
y k =Cx k +Du k +Mh k +v k ；
wherein the subscriptkRepresenting a discrete-state aircraft engine modelkThe data acquisition at the time of the next time,xis that the state variable vector of the aircraft engine comprises the rotating speed of a low-pressure rotorN f And high pressure rotor speedN c ，uIs that the control quantity, i.e. the input deviation, includes the fuel quantityW F And area of the rear nozzleA 8 ，yThe output quantity is measured, namely the output of an actual model comprises high-low pressure rotating speed and total temperature and total pressure of an outlet of a fanT 22 、 P 22 Total temperature and total pressure of outlet of gas compressorT 3 、P 3 Total pressure at the outlet of the combustion chamberP 4 And total pressure of total temperature at outlet of high-pressure turbineT 45 、P 45 And the vector h represents the health parameters fan efficiency and flow of the aircraft engine requiring estimation by the estimatorη f 、W f Compressor efficiency and flowη c 、 W c High pressure turbine efficiency and flowη t 、W t 。w k 、v k Respectively a noise matrix of the system and a measurement noise matrix;
the matrix A, B, C, D, L, M is the engine linear model coefficients, formed by vectorshThe represented health parameters are unknown inputs to the system, which can be viewed as a set of biases, so the baseline model and turbine health estimator can be combined into the form:
2. when the estimator is started, linear combination of all health parameters h, namely model tone, is constructedInteger parameter vectorq=V*hWhereinV * Is a transformation matrix for constructing the adjustment parameter vector, the health parameter vectorh=V *-1 qTherefore, only need to obtainV * Obtaining the estimated value of the health parameter;
3. constructing an aircraft engine model to adapt to the estimation of the turbine health parameters, namely modifying the model into the following form:
4. constructing an estimator model by which the predicted estimator output can be obtained, wherein
P ∞ =A xq P ∞ A xq T -A xq P ∞ C xq T （C xq P ∞ C xq T +R） -1 C xq P ∞ C xq T +Q xq ，
K ∞ =P ∞ C xq T （C xq P ∞ C xq T +R） -1 ；
5. The estimated value is according toK ∞ 、P ∞ The calculation results in that,
then the sum of the squared errors can be calculated as
；
6. In each iteration, the variance of the sum of squared errors with respect to the previous iteration is evaluated to determine if the output error has achieved convergence within a specified range, and if so, 7 is skipped and 8 is continued, otherwise 8 updates are madeV * ；
7. Updating by non-linear least squares functionV * Require
=1 and return 2;
8. return after convergenceV * And ending the flow in the turbine health estimator;
9. output by a turbine health estimatorV * Substituting into formulah=V *-1 qCalculating to obtain an estimate of a turbine health parameterh={η f ，W f ，η c ，W c ，η t ，W t }Respectively representing the efficiency and flow of the fan, the compressor and the high-pressure turbine, and correcting the health parameters of the onboard baseline model of the engineObtaining actual baseline model outputy＇={N c ＇ ,m ＇ HPC ,T 3 ＇ , m ＇ core , T 4 ＇ }，Respectively representing the high-pressure rotor speed after degradationN c ＇ ，Induced air flow of compressorm ＇ HPC ，Bleed air temperature of compressorT 3 ＇ ，Turbine inlet flowm ＇ core ,And turbine inlet temperatureT 4 ＇ ；
10. By the mode of FIG. 3, a calculation module is established by considering factors such as the outer wall of a casing, the inner wall of the casing, the surface of a wheel disc and the like, a blade tip clearance reference model is established, and the actual baseline model is outputy＇={N c ＇ ,m ＇ HPC ,T 3 ＇ , m ＇ core , T 4 ＇ }As input to the tip clearance reference model, for the convenience of differential definition
={N c ＇ ,m ＇ HPC ,T 3 ＇ , m ＇ core , T 4 ＇ }So that
Obtaining tip clearance estimate output for input to a tip clearance model
；
11. Building a deep convolutional neural networkA network, which is a three-layer hidden layer structure; the learning rate is set to 0.01, the training batch is 2000, the iteration is 300 times, and the output layer is the output parameter tip clearance correction valuer'The hidden layer nodes are adjusted according to the training effect of the model, the initial value is 8, and the input is the output of the actual baseline modely＇={N c ＇ ,m ＇ HPC ,T 3 ＇ , m ＇ core , T 4 ＇ }For convenience of distinction, the input defining the neural network model is x＇={N c ＇ ,m ＇ HPC ,T 3 ＇ , m ＇ core , T 4 ＇ }Compensating the blade tip clearance of the engine nonlinear model through a deep convolution network to obtain a blade tip clearance correction valuer'；
12. Obtaining high-precision input parameters by using tip clearance perception modelr'Calculating according to thermal deformation modules of blades and casings in the built blade tip clearance model sensing module, and adding a blade tip clearance correction value obtained through a neural network model
Finally obtaining accurate blade tip clearance estimated value
；
13. A bench test and evaluation platform shown in FIG. 1 is built to verify the accuracy of the parameter perception model of the aero-engine built by the invention on engine parameter perception, the data analysis and data storage module is used for storing the test run data, and the accuracy of the parameter perception model of the aero-engine is evaluated in a deviation form after the engine test run experiment is completed once.
Example 2
The method comprises the steps of establishing a high-precision high-real-time engine nonlinear mathematical model by a component method in consideration of dynamic uncertainty factors, so that the performance evaluation is carried out on test data of a complete machine bench of the engine to realize the self-correction of the component characteristics of an airborne model, a high-precision baseline airborne model is obtained, and the high-precision requirement of a tip clearance model on input parameters is met;
establishing an on-line estimation of the blade tip clearance based on an airborne adaptive model, wherein the problem to be solved is how to build an accurate reference model and the airborne adaptive model matched with the reference model. The reference model in the invention is built by a Tip Clearance model mentioned in Integrated Turbine Tip Clearance and Gas Turbine Engine Simulation, the model structure of the model is divided into three parts, namely a casing model, a wheel disc model and a blade model, which are the most complete and most accurate models in Tip Clearance estimation so far, the input of the whole model is the transient output of an aeroengine model, namely the high-pressure shaft speed, the compressor induced air flow, the compressor induced air temperature, the Turbine inlet flow and the Turbine inlet temperature, the obtained output is the Tip Clearance estimation value, the thermodynamic deformation model and the rotor dynamic deformation model of each part are respectively calculated by the method, and the Turbine Tip Clearance model is finally obtained by synthesis, but the reference model cannot judge whether the blade is subjected to thermal deformation or creep deformation, and only can sense the blade Tip Clearance model under an ideal condition, Blade tip clearance estimation values when the wheel disc and the casing are not deformed;
estimating the flow rate and the efficiency degradation of the rotor part of the engine in real time according to the measurable output parameter deviation by the state estimator with the optimal parameter adjustment, modifying the airborne baseline model to adapt to the real performance degradation condition of the engine, and solving the problem of low input confidence coefficient of the tip clearance model during degradation; secondly, offline learning is carried out by taking the output parameters of the section parameters of the engine and the health parameters of the turbine as input through a deep convolutional neural network algorithm, the blade tip clearance degradation amount is obtained in a neural network model mode, the estimated value of the blade tip clearance of a reference model is corrected online, the real-time requirement of the onboard precision is met while the precision of an onboard baseline model is improved, and the real-time perception of the blade tip clearance with high confidence level in the whole life cycle of the aero-engine is completed.
The embodiments of the present invention have been described in detail with reference to the drawings, but the present invention is not limited thereto, and various changes can be made within the knowledge of those skilled in the art without departing from the gist of the present invention.
Claims (3)
1. An online sensing method for blade tip clearances in a full life cycle of an aircraft engine is characterized by comprising the following specific steps:
s1, establishing an airborne real-time perception baseline model of the working parameters of the aero-engine;
s11, self-correction of the characteristics of the airborne model component is realized by evaluating the performance of the test data of the whole engine bench test;
s12, establishing a bench test database for evaluation to obtain an aircraft engine airborne baseline model with high confidence level, and simultaneously carrying out linearization on the aircraft engine airborne baseline model to construct a discrete linear time-invariant aircraft engine state space equation as described below
x
k+1
=Ax
k
+Bu
k
+Lh
k
+w
k
y k =Cx k +Du k +Mh k +v k ；
Wherein the subscriptkRepresenting a discrete-state aircraft engine modelkThe data acquisition at the time of the next time,xis that the state variable vector of the aircraft engine comprises the rotating speed of a low-pressure rotorN f And high pressure rotor speedN c ，uIs that the control quantity, i.e. the input deviation, includes the fuel quantityW F Area of the nozzle tipA 8 ；yIs that the measured output quantity, i.e. the actual model output, includes high and low pressure rotation speedsN c 、N f (ii) a Total temperature and pressure at fan outletT 22 、P 22 Total temperature and total pressure at outlet of gas compressorT 3 、P 3 Total pressure at outlet of combustion chamberP 4 And total pressure of total temperature at outlet of high-pressure turbineT 45 、P 45 Vector of motionhRepresenting fan efficiency and flow as health parameters of an aircraft engine requiring estimator estimationη f 、W f (ii) a Compressor efficiency and flow
η c 、W c High pressure turbine efficiency and flowη t 、W t ；w k 、v k Respectively a noise matrix of the system and a measurement noise matrix;
the matrix A, B, C, D, L, M is the engine linear model coefficients; the health parameter represented by the vector h is an unknown input to the system;
s2, constructing a thermodynamic and rotor dynamics equation, and calculating radial deformation of a casing, a wheel disc and blades to obtain a mathematical model representing the change of the tip clearance of the turbine in real time, namely an aeroengine tip clearance sensing model;
s3, building a turbine health parameter estimator and a blade tip clearance degradation estimator; supplementing the output of the tip clearance reference model through a neural network and finally establishing an aero-engine full life cycle tip clearance perception model based on model prediction;
the baseline model and turbine health estimator are combined into the form:
s301, starting a turbine health estimator, and constructing all health parametershLinear combination of (2), i.e. model adjustment parameter vectorq=V* hWhereinV * Is a transformation matrix for constructing the adjustment parameter vector, the health parameter vectorh=V *-1 qTo obtainV * Obtaining the estimated value of the health parameter;
s302, constructing an aircraft engine model to be matched with the estimation of the health parameters of the turbine, namely changing the model in the S3 into the following form:
s303, constructing an estimator model, and obtaining the predicted output of the estimator through the estimator model, wherein
P ∞ =A xq P ∞ A xq T -A xq P ∞ C xq T （C xq P ∞ C xq T +R） -1 C xq P ∞ C xq T +Q xq ，
K
∞
=P
∞
C
xq
T
（C
xq
P
∞
C
xq
T
+R）
-1
；
S304, evaluation according to S303K ∞ 、P ∞ The calculation of (a) yields that,
then the sum of the squared errors can be calculated as
；
S305, in each iteration, evaluating the change of the sum of squared errors relative to the previous iteration to determine whether the output error achieves convergence within a specified range, if so, skipping S306 and continuing to execute S307, otherwise, updating S306V * ；
S306, updating through a nonlinear least square functionV * Require
=1 and return to S301;
s307, returning after convergenceV * And ending the flow in the turbine health estimator;
s308, output by the turbine health estimatorV * Substituting into formulah=V *-1 qCalculating to obtain an estimate of a turbine health parameterh= {η f ，W f ，η c ，W c ，η t ，W t }，Respectively representing the efficiency and flow of a fan, an air compressor and a high-pressure turbine, correcting the health parameters of an onboard baseline model of the engine, and simultaneously obtaining the output of an actual baseline modely＇={N c ＇ ,m ＇ HPC ,T 3 ＇ , m ＇ core , T 4 ＇ }，Respectively representing the high-pressure rotor speed after degradationN c ＇ ，Induced air flow of compressorm ＇ HPC ，Bleed air temperature of compressorT 3 ＇ ，Turbine inlet flowm ＇ core ,And turbine inlet temperatureT 4 ＇ ；
S309, establishing a calculation module by considering the surface factors of the outer wall, the inner wall and the wheel disc of the casing, establishing a blade tip clearance reference model, and outputting the reference model through an actual baseline modely＇={N c ＇ ,m ＇ HPC ,T 3 ＇ , m ＇ core , T 4 ＇ }As input of the tip clearance reference model, the definition is convenient for distinguishing
={N c ＇ ,m ＇ HPC ,T 3 ＇ , m ＇ core , T 4 ＇ }So that
Obtaining tip clearance estimate output for input to a tip clearance model
；
S310, building a deep convolution neural network, setting the learning rate to be 0.01, training the batch to be 2000, iterating for 300 times, setting the output layer to be the output parameter tip clearance correction value, adjusting the number of nodes of the hidden layer according to the training effect of the model, setting the initial value to be 8, and inputting the input value to be the output of the actual baseline modely＇={N c ＇ ,m ＇ HPC ,T 3 ＇ , m ＇ core , T 4 ＇ }Definition ofx＇As inputs to neural network models，x＇= {N c ＇ ,m ＇ HPC ,T 3 ＇ , m ＇ core , T 4 ＇ }Compensating the blade tip clearance of the engine nonlinear model through a deep convolution network to obtain a blade tip clearance correction valuer'；
S311, obtaining high-precision input parameters by using tip clearance perception modelr'Calculating according to thermal deformation modules of blades and casings in the built blade tip clearance model sensing module, and adding a blade tip clearance correction value obtained through a neural network model
Finally obtaining accurate online blade tip clearance estimation value
；
S312, a bench test and an evaluation platform are built to verify the accuracy of the parameter perception model in sensing the engine parameters, test run data are stored through a data analysis and data storage module, and after an engine test run experiment is completed once, the accuracy of the parameter perception model of the aero-engine is evaluated in a deviation mode.
2. The method for sensing the full-life-cycle tip clearance of the aero-engine as claimed in claim 1, wherein the aero-engine operating parameter airborne real-time sensing baseline model established in the step S1 comprises an airborne real-time sensing module, a data sensing module and a data management and visualization module.
3. The method for on-line sensing of the full-life-cycle blade tip clearance of the aircraft engine as claimed in claim 1, wherein in S2, geometric features of turbine blade tip clearance key parts and material characteristics changing with temperature are considered, and thermodynamic and rotor dynamics equations are constructed by utilizing a one-dimensional finite difference method and a concentrated mass method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210732109.9A CN114818205B (en) | 2022-06-27 | 2022-06-27 | An online sensing method for tip clearance in the whole life cycle of aero-engine |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210732109.9A CN114818205B (en) | 2022-06-27 | 2022-06-27 | An online sensing method for tip clearance in the whole life cycle of aero-engine |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114818205A CN114818205A (en) | 2022-07-29 |
| CN114818205B true CN114818205B (en) | 2022-09-09 |
Family
ID=82521149
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210732109.9A Active CN114818205B (en) | 2022-06-27 | 2022-06-27 | An online sensing method for tip clearance in the whole life cycle of aero-engine |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114818205B (en) |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115577632B (en) * | 2022-10-14 | 2025-10-28 | 南京航空航天大学 | A method for rapid estimation and control of turbine tip clearance after performance degradation based on machine learning |
| CN115688609A (en) * | 2022-12-19 | 2023-02-03 | 西北工业大学 | Intelligent thrust prediction and real-time early warning method for aircraft engine |
| CN118332773B (en) * | 2024-03-26 | 2024-12-06 | 南京航空航天大学 | Turbine disc temperature calculation method considering engine performance degradation |
| CN118532238B (en) * | 2024-05-16 | 2025-09-05 | 南京航空航天大学 | Real-time sensing method and device for aircraft engine turbine tip clearance |
| CN119511709B (en) * | 2024-11-12 | 2025-05-06 | 南京航空航天大学 | Method and device for self-correcting input parameters of aircraft engine onboard adaptive model |
| CN120027087B (en) * | 2025-04-22 | 2025-06-27 | 西北工业大学 | Test method for verifying influence of compressor blade tip clearance on performance |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104697798A (en) * | 2015-03-05 | 2015-06-10 | 西安交通大学 | System and method for three-dimensional characterization of blade tip gaps and dynamic detection of optical fibers of aero-engine |
| CN110219736A (en) * | 2019-06-19 | 2019-09-10 | 南京航空航天大学 | Aero-engine Direct Thrust Control Strategy based on Nonlinear Model Predictive Control |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7455495B2 (en) * | 2005-08-16 | 2008-11-25 | United Technologies Corporation | Systems and methods for monitoring thermal growth and controlling clearances, and maintaining health of turbo machinery applications |
| US8315741B2 (en) * | 2009-09-02 | 2012-11-20 | United Technologies Corporation | High fidelity integrated heat transfer and clearance in component-level dynamic turbine system control |
| CN102855349B (en) * | 2012-08-06 | 2015-07-01 | 南京航空航天大学 | Quick prototype design method and platform for gas path fault diagnosis for aeroengine |
| CN108062428B (en) * | 2017-10-30 | 2021-07-09 | 南京航空航天大学 | Method and system for on-line component fault diagnosis of turbofan engine |
| CN113267289B (en) * | 2021-04-16 | 2022-08-16 | 上海交通大学 | Array type flexible piezoelectric sensor for aircraft engine and preparation method thereof |
| CN113642271B (en) * | 2021-08-20 | 2022-07-01 | 南京航空航天大学 | Model-based aeroengine performance recovery control method and device |
-
2022
- 2022-06-27 CN CN202210732109.9A patent/CN114818205B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104697798A (en) * | 2015-03-05 | 2015-06-10 | 西安交通大学 | System and method for three-dimensional characterization of blade tip gaps and dynamic detection of optical fibers of aero-engine |
| CN110219736A (en) * | 2019-06-19 | 2019-09-10 | 南京航空航天大学 | Aero-engine Direct Thrust Control Strategy based on Nonlinear Model Predictive Control |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114818205A (en) | 2022-07-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114818205B (en) | An online sensing method for tip clearance in the whole life cycle of aero-engine | |
| CN111914362B (en) | Self-adaptive method for turbofan engine model in research and development stage | |
| CN110502840B (en) | On-line prediction method of aero-engine gas path parameters | |
| CN108647428B (en) | An adaptive component-level simulation model building method for turbofan engine | |
| Kim | A new performance adaptation method for aero gas turbine engines based on large amounts of measured data | |
| CN114154234B (en) | Aeroengine modeling method, system and storage medium | |
| US20140005909A1 (en) | Real time linearization of a component-level gas turbine engine model for model-based control | |
| CN108829928A (en) | A kind of turboshaft engine self-adaptive component grade simulation model construction method | |
| CN116702380A (en) | Aeroengine Performance Degradation Monitoring and Model Correction Method Based on Digital Twin | |
| CN109031951B (en) | On-line establishment of aero-engine state variable model based on exact partial derivatives | |
| CN120008934B (en) | A method for evaluating the stability margin of an aviation gas turbine engine | |
| CN111611743B (en) | Axial-flow compressor characteristic line self-adaption method | |
| CN116822296A (en) | Turbine transition state blade tip clearance estimation method based on long-term and short-term memory neural network | |
| CN113255955A (en) | CNN-RNN-based aero-engine thrust estimation method | |
| CN119511709A (en) | Method and device for self-correcting input parameters of aircraft engine onboard adaptive model | |
| Han et al. | A novel, reduced-order optimization method for nonlinear model correction of turboshaft engines | |
| CN110362960B (en) | Aero-engine system identification method based on multi-cell reduced balanced manifold expansion model | |
| JP5845705B2 (en) | Gas turbine performance estimation device | |
| CN112668162A (en) | Aero-engine modeling method based on inertia sliding mode | |
| CN116796642A (en) | A digital test method for aeroengine testing | |
| CN119337772A (en) | A method and device for adaptively correcting performance of a wide range of aircraft engine components | |
| CN118568869B (en) | General calculation model of mainstream parameters of turbofan engine and individual adaptive correction method | |
| CN113447271A (en) | Online monitoring method for gas circuit of aero-engine based on corrected partial derivative | |
| CN116756866A (en) | Real-time calculation method and system for engine dynamic performance based on bench test data | |
| Zhang et al. | Research on aero-engine physics-based model correction method based on mechanism fusion residual |
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 |