CN103487723A - Electric system fault diagnosis method and system - Google Patents

Abstract

The present invention provides a power system fault diagnosis method and system. The method includes the following steps: receiving alarm information generated after a power system fault occurs, the alarm information including timing information; determining the fault area according to the alarm information; Each faulty element is modeled according to the logical relationship between each faulty element in the faulty area and the corresponding protection and circuit breaker action, and the weighted fuzzy temporal Petri net fault diagnosis model of the faulty element is established; according to the weighted fuzzy temporal Petri net The inference analysis of the fault diagnosis model constructs the corresponding matrix for inference operation and diagnoses the faulty components. The power system fault diagnosis method and system of the present invention effectively improve the fault tolerance and accuracy of fault diagnosis in the power system.

Description

Power system failure diagnostic method and system

Technical field

The present invention relates to the power system security processing technology field, particularly relate to a kind of power system failure diagnostic method and a kind of power system fault diagnosis.

Background technology

Power system failure diagnostic is exactly that the warning information produced after utilizing fault to occur is located fault element quickly and accurately, for the dispatcher provides decision support, thereby rapidly removing faults, recovery system, to normal operating condition, improves the operation stability of system.Done a large amount of research work both at home and abroad on the power system failure diagnostic direction, proposed the different fault diagnosis method, wherein simple, the physical significance of model clearly Petri net method for diagnosing faults be widely applied.

Traditional application Petri net carries out in the method for power system failure diagnostic, for protection and isolating switch, the uncertain problems such as tripping or malfunction may occur, and improves the fault-tolerance of fault diagnosis by setting up the Fuzzy Petri Net inference pattern.Yet said method only utilizes the protection that receives and the action message of isolating switch, when complex fault occurring and follow protection and isolating switch tripping or malfunction and warning information self distorted, may cause obtaining correct diagnostic result.

Summary of the invention

Based on this, the invention provides a kind of power system failure diagnostic method and system, can improve the accuracy of fault diagnosis.

For achieving the above object, the present invention adopts following technical scheme:

A kind of power system failure diagnostic method comprises the following steps:

Receive after electric power system fault occurs the warning information produced, described warning information comprises time sequence information;

Determine fault zone according to described warning information;

To each fault element modeling, set up the Weighted Fuzzy temporal Petri nets fault diagnosis model of fault element according to the logical relation between each fault element in described fault zone and corresponding protection, isolating switch action;

According to the rational analysis of described Weighted Fuzzy temporal Petri nets fault diagnosis model, construct corresponding matrix and carry out the reasoning computing, diagnose the element that is out of order.

A kind of power system fault diagnosis comprises:

The warning information receiver module, for receiving after electric power system fault occurs the warning information produced, described warning information comprises time sequence information;

The fault zone determination module, for determining fault zone according to described warning information;

Model building module, to each fault element modeling, set up the Weighted Fuzzy temporal Petri nets fault diagnosis model of fault element for the logical relation between each fault element according to described fault zone and corresponding protection, isolating switch action;

The reasoning computing module, for the rational analysis according to described Weighted Fuzzy temporal Petri nets fault diagnosis model, construct corresponding matrix and carry out the reasoning computing, diagnoses the element that is out of order.

By above scheme, can be found out; a kind of power system failure diagnostic method of the present invention and system; the time delay binding feature of existence between protection and isolating switch action and the possibility of protection and isolating switch malfunction and tripping have been considered; propose a kind of electric system Weighted Fuzzy temporal Petri nets fault diagnosis model that can take into account this time delay constraint on the basis of existing fault diagnosis model, and carried out fault diagnosis according to this Weighted Fuzzy temporal Petri nets fault diagnosis model.Because fault diagnosis model of the present invention can be processed time delay restricted problem, therefore effectively improved fault-tolerance and the accuracy of fault diagnosis in the electric system.

The accompanying drawing explanation

The schematic flow sheet that Fig. 1 is a kind of power system failure diagnostic method in the embodiment of the present invention;

Fig. 2 is the fault diagnosis model schematic flow sheet based on the Weighted Fuzzy temporal Petri nets in the embodiment of the present invention;

The Weighted Fuzzy temporal Petri nets fault diagnosis model schematic diagram that Fig. 3 is embodiment of the present invention median generatrix;

The Weighted Fuzzy temporal Petri nets fault diagnosis model schematic diagram that Fig. 4 is circuit in the embodiment of the present invention;

Fig. 5 is IEEE New England 10 machine 39 node system schematic diagram in the embodiment of the present invention;

The structural representation that Fig. 6 is a kind of power system fault diagnosis in the embodiment of the present invention.

Embodiment

Shown in Figure 1, a kind of power system failure diagnostic method comprises the following steps:

Step S101, receive after electric power system fault occurs the warning information produced, and described warning information comprises time sequence information, then enters step S102.

Step S102, determine fault zone according to described warning information, then enters step S103.

As an embodiment preferably, the process of described definite fault zone specifically can comprise: adopt breadth-first search to determine described fault zone.

Step S103, to each fault element modeling, set up the Weighted Fuzzy temporal Petri nets fault diagnosis model of fault element according to the logical relation between each fault element in described fault zone and corresponding protection, isolating switch action, then enters step S104.

Step S104, according to the rational analysis of described Weighted Fuzzy temporal Petri nets fault diagnosis model, construct corresponding matrix and carry out the reasoning computing, diagnoses the element that is out of order.

As an embodiment preferably, after element is out of order in diagnosis, can also comprise the steps:

Step S105, carry out backward reasoning to the fault element of diagnosing out, draws malfunction and the tripping situation of protection and isolating switch.

1), determine power supply interrupted district (being fault zone) in another embodiment, fault diagnosis model of the present invention can be divided into following four layers:; 2), set up the element fault diagnostic model; 3), the reasoning based on matrix computations; 4), protection and isolating switch action evaluation.Detailed process is shown in Figure 2:

1, the isolating switch displacement information provided according to network topology structure and data acquisition and surveillance (supervisory control and data acquisition, SCADA), adopt BFS (Breadth First Search) to determine fault zone.If fault zone only comprises an element, this element is fault element; If comprise two or more elements in fault zone, each element is wherein set up respectively to Weighted Fuzzy temporal Petri nets network model and carry out the failure judgement element;

2., according to the logical relation between fault element and corresponding protection, isolating switch action, build the Weighted Fuzzy temporal Petri nets diagnostic model of element, then merged according to each interelement topological relation;

3, the Weighted Fuzzy temporal Petri nets diagnostic model based on developed, realize rational analysis by matrix operation, the final diagnosis element that is out of order;

4, the fault element of diagnosing out is taked to backward reasoning, with malfunction and the tripping situation of judgement protection and isolating switch.

Below troubleshooting step of the present invention is elaborated.

One, the mathematical description of Weighted Fuzzy temporal Petri nets:

1), definition Weighted Fuzzy temporal Petri nets is 11 tuples:

S _WFTPN={P,T,I,O,A _cc,ΔT _min,ΔT _max,U,T _h,W,M} (1)

In formula: P={p ₁, p ₂..., p _nby storehouse, collected; T={t ₁, t ₂..., t _mit is the transition collection; For characterizing inference rule; I:P → T is the mapping of reflection transition that storehouse is arrived; I=[δ _ij] be n * m matrix; Work as p _it _jinput (have p _ito t _jdirected arc) time δ _ij=1; Otherwise δ _ij=0; O:T → P reflection be transitted towards storehouse mapping; O=[γ _ij] be m * n matrix; Work as p _jt _ioutput (have t _ito p _jdirected arc) time γ _ij=1; Otherwise γ _ij=0; A _cc=[a _ij] be n * n matrix; Characterize general storehouse arrive the purpose storehouse path; Work as p _istorehouse institute path through p _jthe time a _ij=1; Otherwise a _ij=0; Δ T _min=[Δ τ _1min, Δ τ _2min..., Δ τ _nmin] be storehouse with the minimum time delay constraint of preposition transition; Δ T _max=[Δ τ _1max, Δ τ _2max,, Δ τ _nmax] be storehouse with the constraint of the maximum delay of preposition transition; If Δ τ _min=Δ τ _max=0; Transition moment activation; U=[μ ₁, μ ₂..., μ _m] for the degree of confidence vector of transition; If for any j, μ is arranged _j=1; Model is the simple Petri net that does not contain fuzzy variable; T _h=[λ ₁, λ ₂..., λ _m] for the igniting threshold vector of transition; W=diag (w ₁, w ₂..., w _n) for inputting the weight matrix of arc; The reflection precondition is to regular influence degree; The event type that its value characterizes to storehouse is relevant; M=[α (p ₁), α (p ₂) ..., α (p _n)] be storehouse institute degree of confidence vector; α (p _i) p of library representation institute _idegree of confidence.

The Petri net can pass through the node P={p of storehouse institute ₁, p ₂..., p _n, transition node T={t ₁, t ₂..., t _mand directed arc mean.Directed arc comprises input arc and output arc two classes, and the input arc is by storehouse transition pointed, and the output arc points to storehouse institute by transition.Wherein, storehouse circle used " zero " means, vertical line for transition " | " expression.The structure of Weighted Fuzzy Petri Net and the maximum difference of simple Petri net be to consider the input arc weights, storehouse degree of confidence, the degree of confidence of transition, the transition process of probable value etc.In addition, the probable value of the event of institute's value representation storehouse, Weighted Fuzzy Petri Net storehouse representative, even transition are activated, input magazine is worth and remains unchanged, library representation institute degree of confidence still, and can be along with transition produce storehouse institute.Consider that on the basis of Weighted Fuzzy Petri Net model the time delay between the institute of storehouse retrains, and is Weighted Fuzzy temporal Petri nets model.This model can be processed time delay restricted problem, can improve fault-tolerance and the accuracy of model.

2), the rational analysis of transition temporal constraint, forward direction collection and the backward collection of transition t ∈ T are defined as respectively ^.t={p| δ _pt=1}; t ^.={ p| γ _tp=1}.

Component library action delay constraint be to being connected to same transition t both sides Liang Ge storehouse institutes (successive relay trip) ^.t and t ^.the time delay interval

with

carry out interconnection constraint, i.e. Δ T _t=[Δ τ _tmin, Δ τ _tmax], be embodied in these two component libraries the connection transition on.

Forward inference: known

with Δ T _t=[Δ τ _tmin, Δ τ _tmax], can obtain component library institute ^.the follow-up component library of t action delay constraint:

$T_{t^{.}}=T_{t_{.}}+\mathrm{\Δ}{T}_t=[{\mathrm{\τ}}_{t\min _{.}},{\mathrm{\τ}}_{t\max _{.}}]+[{\mathrm{\Δ\τ}}_{t\min },{\mathrm{\Δ\τ}}_{t\max }]=[{\mathrm{\τ}}_{t\min _{.}}+{\mathrm{\Δ\τ}}_{t\min _{.}},{\mathrm{\τ}}_{t\max _{.}}+\mathrm{\Δ}{\mathrm{\τ}}_{t\max }]---\left(2\right)$

Backward reasoning: known

with Δ T _t=[Δ τ _tmin, Δ τ _tmax], can obtain the t of component library institute ^.forerunner's component library action delay constraint:

$T_{t_{.}}=T_{t_{.}}-\mathrm{\Δ}{T}_t=[{\mathrm{\τ}}_{t^{.}\min },{\mathrm{\τ}}_{t^{.}\max }]-[\mathrm{\Δ}{\mathrm{\τ}}_{t\min },{\mathrm{\Δ\τ}}_{t\max }]=[{\mathrm{\τ}}_{t^{.}\min }-{\mathrm{\Δ\τ}}_{t\max },{\mathrm{\τ}}_{t^{.}\max }-{\mathrm{\Δ\τ}}_{t\min }]---\left(3\right)$

3), for Weighted Fuzzy Petri Net, its fuzzy reasoning depends on fuzzy rule set R={R ₁, R ₂..., R _m}:

R _i:Ifd _jThend _k(CF=μ _i)(i=1,2,…,m) (4)

In formula: d _jand d _kfor the proposition containing fuzzy variable, its confidence alpha (d _j) and α (d _k) value in interval [0,1], in the Petri net, be presented as storehouse degree of confidence; μ _i∈ [0,1] shows regular R _ithe degree of confidence value, it is presented as the degree of confidence of transition in Petri net.

In the reasoning process of Weighted Fuzzy temporal Petri nets, need to consider the time delay constraint simultaneously.Suppose that A, B and C are m * n matrix, and D is m * q matrix, E is q * n matrix; Definition: (1) addition operator ⊕: C=A ⊕ B, c _ij=max (a _ij, b _ij); (2) comparison operator

: work as a _ij>=b _ijthe time c _ij=1, otherwise c _ij=0; (3) directly take advantage of c of operator e:C=AeB _ij=a _ijb _ij; (4) multiplication operator

:

(5) matrix multiplication g:C=DgE,

The reasoning process of Weighted Fuzzy temporal Petri nets is for obtaining a stable network state, and the value of storehouse institute degree of confidence matrix M is no longer carried out with iteration and the state that changes.At first retrain alarm is screened by time delay.Suppose that the k time iteration obtains the degree of confidence matrix M ^k, obtain the degree of confidence matrix M the k+1 time ^k+1reasoning process as follows:

Step1: definition storehouse institute path matrix is A _cc.According to temporal constraint, the minimum cumulative delay constraint matrix Σ Δ T that storehouse institute path is corresponding is obtained in rational analysis _min, and cumulative maximum time delay constraint matrix Σ Δ T _max, be all n * 1 vector:

ΣΔT _min=(A _ccg(ΔT _min) ^T) ^T (5)

ΣΔT _max=(A _ccg（ΔT _max） ^T) ^T (6)

Step2: the sequential consistance of check warning information, screen on this basis alarm.Definition F=[f ₁, f ₂..., f _n] be the time delay constraint qualification vector of protection and isolating switch, element wherein characterizes time delay that whether corresponding storehouse meets that model bank is concentrated constraint, f _i=1 and 0 (i=1; 2; N) mean respectively to meet and do not meet constraint.Delayed data set Δ T by protection and isolating switch in warning information _mesminwith Δ T _mesmaxcompare with the time delay constraint, shown in (7), can obtain:

Step3: given original state M ⁰, make k=0;

Step4: the weights that calculate the input arc:

W _inarc=WgI (8)

Step5: the synthetic input confidence level of calculating transition:

E ^k=M ^kgW _inarc (9)

Step6: synthetic input confidence level and the threshold value of transition are compared; Be met the transition set of activation condition:

Step7: the synthetic input confidence level that calculating can make transition activate:

H ^k=E ^keG ^k (11)

Step8: the M that calculates the k+1 time reckoning of storehouse institute ^k+1:

Step9: if M ^k+1=M ^k, the degree of confidence matrix of Petri net is stable, otherwise makes k=k+1, returns to Step4.

Two, Weighted Fuzzy temporal Petri nets fault diagnosis model, after the element generation permanent fault in electric system, relevant protection and isolating switch can move, and finally form one or more power supply interrupted districts, and fault element is certainly among power supply interrupted district.The fault diagnosis model below developed is exactly for element included in power supply interrupted district.

The IEEE39 node system of take is set up the Weighted Fuzzy temporal Petri nets fault diagnosis model of bus and circuit as example.Suppose that every circuit two ends have configured main protection, nearly back-up protection and back-up protection far away.Back-up protection on the main protection of having supposed every bus bar configuration, the circuit that is connected.Mean respectively bus, circuit, main protection, nearly back-up protection and back-up protection far away with B, L, m, p and s.For example, CB _(4)-14mean the isolating switch of circuit L4-14 near bus B4 side, R _(4)-14mmean the main protection of circuit L4-14 near bus B4 side, the rest may be inferred.

1), the protection sequence of movement is followed successively by main protection, nearly back-up protection and back-up protection far away.Defining them is respectively with respect to the time delay of fault moment

with

the isolating switch that defines all kinds of protection correspondences is respectively with respect to the time delay of operating time of protection $\mathrm{\Δ}{T}_{\mathrm{mc}}\∈[{\mathrm{\Δ\τ}}_{\mathrm{mc}}^{\min },{\mathrm{\Δ\τ}}_{\mathrm{mc}}^{\max }],$ $\mathrm{\Δ}{T}_{\mathrm{pc}}\∈[{\mathrm{\Δ\τ}}_{\mathrm{pc}}^{\min },{\mathrm{\Δ\τ}}_{\mathrm{pc}}^{\max }]$ With $\mathrm{\Δ}{T}_{\mathrm{sc}}\∈[{\mathrm{\Δ\τ}}_{\mathrm{sc}}^{\min },{\mathrm{\Δ\τ}}_{\mathrm{sc}}^{\max }].$

2), basic ideas that bus is set up to fault diagnosis model are: the model of its each closure main protection of model, back-up protection, then build bus resultant fault diagnostic model.Similarly, the basic ideas of circuit being set up to fault diagnosis model are: all kinds of protection models at its two ends of model, then build circuit resultant fault diagnostic model.

The bus B14 of take in Fig. 5 is example, and it has configured main protection R _b14mand the back-up protection R far away of connection line _(4)-14s, R _(13)-14sand R _{14-(15) s}, obtain the Weighted Fuzzy temporal Petri nets fault diagnosis unified model of this bus as shown in Figure 3.Take circuit L4-14 as example, and it has configured main protection R _(4)-14mand R _{4-(14) m}, nearly back-up protection R _(4)-14pand R _{4-(14) p}, back-up protection R far away _(3)-4s, R _{4-(5) s}, R _{14-(15) s}and R _(13)-14s, obtain the Weighted Fuzzy temporal Petri nets fault diagnosis unified model of this circuit as shown in Figure 4.

3) parameter determination method, after having built fault diagnosis model, next step just is based on the reasoning process of matrix operation, and it is consistent with the fuzzy reasoning based on fuzzy rule.Wherein, relevant parameters is determined as follows:

1. the arc weights are inputted in transition

For each end of circuit, consider main protection, nearly back-up protection and the back-up protection far away Different Effects to circuit, with and with the coordinating of respective circuit breakers.Like this, input arcs for protection storehouse institute, isolating switch storehouse institute to two of transition and give respectively weights, as shown in table 1.

The protection that table 1 is given and isolating switch weights

2. the degree of confidence initial value is given

To protection and the corresponding storehouse of isolating switch alarm institute, given its degree of confidence initial value, as shown in table 2.

The degree of confidence of the protection that table 2 is given and isolating switch alarm

Consider that the warning information received may be mistake or incomplete, for the protection in alarm and isolating switch degree of confidence all are not set to 0.2.From the fault-tolerance angle, the degree of confidence that transition are set is 0.95, and the transition threshold value is 0.2, and transition storehouse institute degree of confidence is 1, and transition transition threshold value is 0, and transition transition time delay is 0.

3. time delay constraint

Definition main protection, nearly back-up protection, back-up protection far away, and the time delay of corresponding isolating switch constraint is respectively Δ T _mr∈ [10,40], Δ T _pr∈ [300,500], Δ T _sr∈ [600,1100], Δ T _mc∈ [40,60], Δ T _pc∈ [20,40], Δ T _sc∈ [40,100], unit is ms.

The fault diagnosis model developed has been considered between element fault and protection action, has been protected the time delay between action and circuit breaker trip to retrain; and by model can find out isolating switch and the protection storehouse be all by transition directly arrive bus and line library, like this isolating switch storehouse Δ T _minwith Δ T _maxtime delay constraint constantly should occur with respect to fault for what forward inference was tried to achieve in assignment.

Three, the adaptability of fault diagnosis model and fault-tolerance:

1), the adaptability in the network topology change situation.For example, when the existing fault diagnosis model based on Petri net changes (have newly-increased circuit or existing line out of service) in NETWORK STRUCTURE PRESERVING POWER SYSTEM, often need to rebuild the Petri pessimistic concurrency control, workload is large.And the Weighted Fuzzy temporal Petri nets model in the present invention; setting up bus main protection, back-up protection; and, after the submodel of power line main protection, nearly back-up protection, back-up protection far away, then carry out simple the fusion and form integrated diagnosis model, simple for structurely understand.When having newly-increased bus or existing bus out of service, for the model built, only need newly-increased or delete the model of associated bus, because model structure and the reasoning process of each bus is similarly, so just the energy rapid pin be made correction to network topology change.For circuit model; situation and bus model are similar; but consider that circuit may also be responsible for the standby of associated bus or circuit/far back-up protection simultaneously; when having newly-increased circuit or existing line out of service; not only need to increase or delete the model of this circuit; also need standby to respective bus bars and circuit/back-up protection far away to be adjusted, increase or delete corresponding protection submodel.The fault diagnosis model developed can adapt to network topology change fast, and counting yield is high.

2), the fault-tolerance in the incorrect situation of warning information.The fault diagnosis model of the temporal based Fuzzy Petri Net of setting up in the embodiment of the present invention is by analyzing the degree of confidence of candidate's fault element; can take into account protection and the malfunction of isolating switch/tripping situation; and warning information is not right-on situation, there is fault-tolerance preferably.In addition, according to configuration relation and the action logic of element, protection, isolating switch, adopt backward reasoning can differentiate the malfunction of protection and isolating switch/tripping situation.

Below; for having protection and likely malfunction or tripping of isolating switch in fault diagnosis; in the alarm upload procedure, also may occur uploading not in time, the uncertain factor such as distortion or loss, the present invention be take the IEEE39 node system shown in Fig. 5 and is verified the fault diagnosis model developed as example.Suppose to receive the warning information with sequential: protection R _b14m(20ms), R _(4)-14s(750ms) and R _{14-(15) s}(371ms) action; Isolating switch CB _(14)-15(73ms), CB _(14)-13(81ms) and CB _(4)-14(87ms) tripping.

1), fault zone identification.Warning information based on received, determine fault zone by BFS (Breadth First Search), and possible fault element is B14 and L4-14.

2), to each element modeling in fault zone.Set up bus B14 as shown in Figure 3, Figure 4 and the Weighted Fuzzy temporal Petri nets fault diagnosis model of circuit L4-14.

3), the rational analysis based on matrix operation.According to the rational analysis of Weighted Fuzzy temporal Petri nets, construct corresponding matrix and carry out the reasoning computing:

1. adopt matrix operation to carry out rational analysis to the Weighted Fuzzy temporal Petri nets model of bus B14:

At first according to the time delay constraint, the alarm of receiving is screened, storehouse institute path matrix, minimum cumulative delay constrained vector and cumulative maximum time delay constrained vector that storehouse institute path is corresponding are respectively:

(annotate: the R of storehouse institute _b14mstorehouse institute path only need count t ₁, t ₃and t ₅in one get final product because three transition are of equal value.)

ΣΔT _min=(A _ccg(ΔT _min) ^T) ^T=[640,640,640,50,50,50,10,600,600,600,0,0,0,0]

ΣΔT _max=(A _ccg(ΔT _max) ^T) ^T=[1200,1200,1200,100,100,100,40,1100,1100,1100,0,0,0,0]

Afterwards, the consistance of check sequential.By the protection in warning information and isolating switch delayed data set Δ T _mesminwith Δ T _mesmaxwith the time delay constraint, compare, the protection and the isolating switch that are met constraint are judged vector (f _i=0 means that sequential is inconsistent):

$F=[{\mathrm{<figure-callout\; id="14m"\; label="f\; B"\; filenames="HDA00003750885800031.png"\; state="\{\{state\}\}">f</figure-callout>}}_B,f_{R_{\left(4\right)-14s}},f_{R_{14-\left(15\right)s}},f_{{\mathrm{CB}}_{\left(14\right)-15}},f_{{\mathrm{CB}}_{\left(14\right)-13}},f_{{\mathrm{CB}}_{\left(4\right)-14}}]$

Like this, the effective warning information after time delay constraint screening is: protection R _b14m(20ms) and R _(4)-14s(750ms) action; Isolating switch CB _(14)-15(73ms), CB _(14)-13(81ms) and CB _(4)-14(87ms) action tripping operation.

The storehouse of Petri pessimistic concurrency control collects:

p={CB _(4)-14,CB _(13)-14,CB _14-(15),CB _4-(14),CB _(14)-13,CB _(14)-15,R _B14m,R _(4)-14s,R _(13)-14s,R _14-(15)s,L4-14,L13-14,L14-15,B14}

The transition collection is:

T={t ₁,t ₂,t ₃,t ₄,t ₅,t ₆,t ₇}

The input matrix of transition:

The output matrix of transition:

Transition degree of confidence vector:

U=[0.95,0.95,0.95,0.95,0.95,0.95,1]

Transition activation threshold vector:

T _h=[0.2,0.2,0.2,0.2,0.2,0.2,0]

Input arc weight vector:

W=diag(0.50,0.50,0.50,0.40,0.40,0.40,0.60,0.50,0.50,0.50,0.33,0.33,0.33,0)

Given storehouse institute original state is:

M ⁰=[0.65,0.20,0.20,0.20,0.85,0.85,0.90,0.70,0.20,0.20,0.00,0.00,0.00,0.00]

Adopting the matrix reasoning process can obtain the degree of confidence that bus B14 breaks down is 0.771.

2. adopt matrix operation to carry out rational analysis to the Weighted Fuzzy temporal Petri nets model of circuit L4-14:

At first according to the time delay constraint, the alarm of receiving is screened, storehouse institute path matrix, minimum cumulative delay constrained vector and cumulative maximum time delay constrained vector that storehouse institute path is corresponding are respectively:

ΣΔT _min=(A _ccg(ΔT _min) ^T) ^T=[640,640,640,640,320,320,50,50,10,10,300,300,600,600,600,600,0,0,0,0,0]ΣΔT _max=(A _ccg(ΔT _max) ^T) ^T=[1200,1200,1200,1200,540,540,100,100,40,40,500,500,1100,1100,1100,1100,0,0,0,0,0]

Afterwards, the consistance of check sequential.By the protection in warning information and isolating switch delayed data set Δ T _mesminwith Δ T _mesmaxwith the time delay constraint, compare, the protection and the isolating switch that are met constraint are judged vector (f _i=0 means that sequential is inconsistent):

$F=[{\mathrm{<figure-callout\; id="14m"\; label="f\; B"\; filenames="HDA00003750885800031.png"\; state="\{\{state\}\}">f</figure-callout>}}_B,f_{R_{\left(4\right)-14s}},f_{R_{14-\left(15\right)s}},f_{{\mathrm{CB}}_{\left(14\right)-15}},f_{{\mathrm{CB}}_{\left(14\right)-13}},f_{{\mathrm{CB}}_{\left(4\right)-14}}]$

Like this, the effective warning information after time delay constraint screening is: protection R _b14m(20ms) and R _(4)-14s(750ms) action; Isolating switch CB _(14)-15(73ms), CB _(14)-13(81ms) and CB _(4)-14(87ms) action tripping operation.

The storehouse of Petri pessimistic concurrency control collects:

p={CB _4-(5),CB _(3)-4,CB _14-(15),CB _(13)-14,CB′ _(4)-14,CB′ _4-(14),CB _(4)-14,CB _4-(14),R _(4)-14m,R _4-(14)m,

R _(4)-14p,R _4-(14)p,R _4-(5)s,R _(3)-4s,R _14-(15)s,R _(13)-14s,p ₁,p ₂,L(4)-14,L4-(14),L4-14}

The transition collection is:

T={t ₁,t ₂,t ₃,t ₄,t ₅,t ₆,t ₇,t ₈,t ₉,t ₁₀,t ₁₁}

The input matrix of transition:

The output matrix of transition:

The time delay constrained vector:

ΔT _min=[640,640,640,640,320,320,50,50,10,10,300,300,600,600,600,600,0,0,0,0,0]

Δ T _max=[1200,1200,1200,1200,540,540,100,100,40,40,500,500,1100,1100,1100,1100,0,0,0,0,0] transition degree of confidence vector:

U=[0.95,0.95,0.95,0.95,0.95,0.95,0.95,0.95,1,1,1]

Transition activation threshold vector:

T _h=[0.2,0.2,0.2,0.2,0.2,0.2,0.2,0.2,0,0,0]

Input arc weight vector:

W=diag(0.50,0.50,0.50,0.50,0.45,0.45,0.40,0.40,0.60,0.60,0.55,0.55,0.50,0.50,0.50,0.50,1,1,0.50,0.50,0

Adopting the matrix reasoning process can obtain the degree of confidence that circuit L4-14 breaks down is 0.219.

4), protection and isolating switch action evaluation

After above-mentioned steps is judged fault element, carry out now backward reasoning.Known after bus B14 breaks down, main protection R _b14maction, trigger isolating switch CB _(14)-15; , CB _(14)-13and CB _4-(14)action tripping operation, and isolating switch CB wherein _4-(14)tripping, therefore corresponding back-up protection R _(4)-14saction tripping isolating switch CB _(4)-14thereby, isolated fault.Like this; Evaluation result is isolating switch CB _4-(14)tripping.

For the fault diagnosis model that checking develops better, in the embodiment of the present invention, also the various faults scene is tested.Table 3 has been listed the diagnostic result to the partial fault scene.Numerical results shows, method proposed by the invention can be processed protection and isolating switch has incorrect operation and the wrong complex fault situation of warning information.

Table 3 numerical testing

Identical with above-mentioned a kind of power system failure diagnostic method, the present invention also provides a kind of power system fault diagnosis, as shown in Figure 6, comprising:

Warning information receiver module 101, for receiving after electric power system fault occurs the warning information produced, described warning information comprises time sequence information;

Fault zone determination module 102, for determining fault zone according to described warning information;

Model building module 103, to each fault element modeling, set up the Weighted Fuzzy temporal Petri nets fault diagnosis model of fault element for the logical relation between each fault element according to described fault zone and corresponding protection, isolating switch action;

Reasoning computing module 104, for the rational analysis according to described Weighted Fuzzy temporal Petri nets fault diagnosis model, construct corresponding matrix and carry out the reasoning computing, diagnoses the element that is out of order.

As an embodiment preferably, described power system fault diagnosis can also comprise:

The action evaluation module, for after element is out of order in diagnosis, carry out backward reasoning to the fault element of diagnosing out, draws malfunction and the tripping situation of protection and isolating switch.

As an embodiment preferably, described fault zone determination module can adopt breadth-first search to determine described fault zone.

Other technical characterictic of above-mentioned a kind of power system fault diagnosis is identical with a kind of power system failure diagnostic method of the present invention, and it will not go into details herein.

By above scheme, can find out; a kind of power system failure diagnostic method of the present invention and system; the time delay binding feature of existence between protection and isolating switch action and the possibility of protection and isolating switch malfunction and tripping have been considered; propose a kind of electric system Weighted Fuzzy temporal Petri nets fault diagnosis model that can take into account this time delay constraint on the basis of existing fault diagnosis model, and carried out fault diagnosis according to this Weighted Fuzzy temporal Petri nets fault diagnosis model.Because fault diagnosis model of the present invention can be processed time delay restricted problem, therefore effectively improved fault-tolerance and the accuracy of fault diagnosis in the electric system.

The above embodiment has only expressed several embodiment of the present invention, and it describes comparatively concrete and detailed, but can not therefore be interpreted as the restriction to the scope of the claims of the present invention.It should be pointed out that for the person of ordinary skill of the art, without departing from the inventive concept of the premise, can also make some distortion and improvement, these all belong to protection scope of the present invention.Therefore, the protection domain of patent of the present invention should be as the criterion with claims.

Claims (6)

1. A power system fault diagnosis method, is characterized in that, comprises the following steps: receiving alarm information generated after a power system failure occurs, and the alarm information includes timing information; determining the fault area according to the alarm information; Each faulty element is modeled according to the logical relationship between each faulty element in the faulty area and the corresponding protection and circuit breaker action, and the weighted fuzzy sequential Petri net fault diagnosis model of the faulty element is established; According to the inference analysis of the weighted fuzzy time-series Petri net fault diagnosis model, a corresponding matrix is constructed for inference operation, and the faulty element is diagnosed. 2. The power system fault diagnosis method according to claim 1, further comprising the step of: Reverse reasoning is carried out on the diagnosed fault components to obtain the malfunction and refusal of protection and circuit breakers. 3. The power system fault diagnosis method according to claim 1 or 2, wherein the process of determining the fault area comprises: using a breadth-first search algorithm to determine the fault area. 4. A power system fault diagnosis system, characterized in that, comprising: The alarm information receiving module is used to receive the alarm information generated after the power system failure occurs, and the alarm information includes timing information; A fault area determining module, configured to determine the fault area according to the alarm information; The model building module is used to model each fault element according to the logical relationship between each fault element in the fault area and the corresponding protection and circuit breaker action, and establishes a weighted fuzzy sequential Petri net fault diagnosis model of the fault element; The reasoning operation module is used for constructing a corresponding matrix to perform reasoning operations according to the reasoning analysis of the weighted fuzzy time-series Petri net fault diagnosis model, and to diagnose faulty components. 5. The power system fault diagnosis system according to claim 4, further comprising: The action evaluation module is used to perform reverse reasoning on the diagnosed faulty components after diagnosing the faulty components, and obtain the malfunction and refusal of protection and circuit breakers. 6. The power system fault diagnosis system according to claim 4 or 5, characterized in that the fault area determination module uses a breadth-first search algorithm to determine the fault area.

Images