CN113204735B - Power grid fault diagnosis method based on random self-regulating pulse nerve P system - Google Patents

Abstract

The invention discloses a power grid fault diagnosis method based on RSSNPS, which comprises the steps of S1, reading SCADA data; s2, determining a power failure area in the power system to be diagnosed; s3, establishing a corresponding fault diagnosis objective function based on the determined power failure area; s4, based on SCADA data, utilizing a random self-adjusting pulse neural P system to perform optimizing solution on the established fault diagnosis objective function to obtain an optimal solution; s5, determining a power grid fault diagnosis result according to the code of the optimal solution. The invention provides a method for improving the global convergence effect of the fault diagnosis process by utilizing the RSSNPS so as to ensure that the optimal solution is accurately searched, and introducing weather factors, self-checking information and three types of self-condition trust factors into a fault diagnosis objective function, thereby improving the fault diagnosis capability of the objective function in the face of complex environmental conditions; the problem of fault diagnosis under abnormal conditions of the protection device and the circuit breaker due to the distortion of fault alarm information caused by disaster weather is effectively solved.

Description

Power grid fault diagnosis method based on random self-regulating pulse nerve P system

Technical Field

The invention belongs to a power grid fault diagnosis method, and particularly relates to a power grid fault diagnosis method based on a Random Self-adjusting pulse-nerve P system (RSSNPS) in disaster-causing weather.

Background

Up to now, many scholars at home and abroad have widely studied and have achieved remarkable results on the fault diagnosis method of the power system. The methods mainly comprise expert systems, optimization technology, artificial neural networks, petri networks, rough set theory, bayesian theory, fault recorder information-based methods and the like. The power system fault diagnosis method based on the optimization technology has strict mathematical logic and strong fault tolerance, and can effectively realize accurate diagnosis of power grid faults by utilizing an optimization algorithm.

Weather factors have a close and indispensible relationship with the failure cause of the power transmission network, and particularly under extreme environmental conditions such as disaster weather, weather related disaster factors such as lightning strokes, typhoons and the like are the main causes of the power transmission network failure. Therefore, in the actual engineering environment, the accuracy of the fault alarm information is reduced due to the interference of other external objective factors, and the protection device can also make an abnormal response.

The power transmission network is subjected to fault diagnosis by an optimization technology, and when the power transmission network is subjected to a fault diagnosis model, the power transmission network is often subjected to a high-dimensional mathematical model, so that a higher requirement is put forward on the convergence effect and the convergence speed of an algorithm. Therefore, when the algorithm is faced with specific problems, if the algorithm suffers from the problems of poor convergence effect and slow convergence speed, the improvement of the algorithm is essential.

Disclosure of Invention

Aiming at the defects in the prior art, the power grid fault diagnosis method based on the random self-regulating pulse neural P system solves the problem that the fault diagnosis result is inaccurate because the influence of weather factors is not considered in the existing power grid fault diagnosis method.

In order to achieve the aim of the invention, the invention adopts the following technical scheme: a power grid fault diagnosis method based on a random self-regulating pulse nerve P system comprises the following steps:

s1, SCADA data is read;

s2, determining a power failure area in the power system to be diagnosed based on the read SCADA data;

s3, establishing a corresponding fault diagnosis objective function based on the determined power failure area;

s4, based on SCADA data, utilizing a random self-adjusting pulse neural P system to perform optimizing solution on the established fault diagnosis objective function to obtain an optimal solution;

s5, determining a power grid fault diagnosis result according to the code of the optimal solution.

Further, the method for determining the outage area in the step S2 specifically includes:

based on SCADA data, performing iterative search on the electric power system to be diagnosed by using a junction line analysis method, and finding out all passive networks in the electric power system to be diagnosed in the iterative search process, namely determining a power failure area;

The method specifically comprises the following sub-steps:

s21, setting an initial value of the search iteration number C to be 1;

s22, numbering each element in the power system to be diagnosed in sequence based on SCADA data, wherein all element numbers form Q _C ；

S23, slave element number setQ is combined _C Any element number is taken and put into the element number subset M _C In (a) and (b);

s24, judging the latest added element number subset M _C Whether the element corresponding to the element number is provided with a closed circuit breaker connected with the element;

if yes, go to step S24;

if not, go to step S25;

s24, assembling the element numbers into a set Q _C Element numbers corresponding to all elements connected with the currently determined closed circuit breaker are added to the element number subset M _C In the step A3, returning to the step;

s25, increasing the search iteration number C by 1;

s26, slave element number set Q _C-1 Removing element number subset M _C-1 All the element numbers in the database are obtained to obtain a new element number set Q _C ；

S27, judging the current element number set Q _C Whether it is empty;

if yes, go to step S28;

if not, returning to the step S23;

s28, list element number subset M ₁ ,...S _C ,...,M _N Determining a power failure area;

the subscript N is the number of element coding subsets obtained in the iterative search process.

Further, the step S3 specifically includes:

s31, establishing a fault hypothesis H consisting of suspicious fault elements, protection devices and circuit breakers in the power failure area according to the topological structure of the power failure area;

s32, constructing a fault diagnosis objective function based on the fault hypothesis.

Further, the method for establishing the fault hypothesis H in step S31 specifically includes:

the power failure setting region contains n _d Suspected faulty elements, n _r Protection device and n _c With circuit breakers, built-up fault hypothesisThe expression is:

X＝[D、R、C]

in the formula,

1≤i≤n _d ，d _i for the actual status of the ith suspect element in the outage area, when d _i At 1, the corresponding suspected failed element has failed, d _i When the value is 0, the corresponding suspicious element is normal and has no fault;

1≤j≤n _r ，r _j for the actual state of the jth protection device in the power failure area, when r _j When the value is 1, the corresponding protection device is operated, r _j When the value is 0, the corresponding protection device does not act;

1≤k≤n _c ，c _k when c is the actual state of the kth breaker in the power failure area _k When 1, the corresponding breaker is tripped, c _k 0, corresponding to the breaker not tripped;

in the step S32, the expression of the fault diagnosis objective function minE (X) is:

min E(X)＝δ _ex E _ex (X)+γ _al E _al (X)+T _p E _tl (X)

in the formula,δ_ex To self-regulate desired trust factor, self-regulate desired trust factor comprising protection means

And self-adjusting desired trust factor of the circuit breaker>

γ _al For self-regulating alarm trust factor, self-regulating alarm trust factor gamma comprising protection means _ral Self-adjusting alarm trust factor gamma with circuit breaker _cal ；

T _p Is a self-adjusting weather trust factor;

E _ex (X) is a function reflecting the malfunction and refusal of the protection device and the circuit breaker, and the calculation formula is as follows:

wherein ,

self-adjusting desired trust factor for the j-th protection means>

Self-adjusting desired trust factor for the kth circuit breaker correspondence,>

for the desired state of the jth protective device, < >>

Is the expected state of the kth circuit breaker;

E _al the (X) is a function reflecting the condition of missing report and false report of alarm information of a protection device and a circuit breaker, and the calculation formula is as follows:

wherein ,

self-adjusting alarm trust factor for the j-th protection device,>

self-adjusting alarm trust factor corresponding to kth circuit breaker,>

for the observation state of the jth protective device, < >>

The observation state of the kth circuit breaker;

E _tl (X) is a function reflecting the matching condition of the fault probability and the real state of the line, and the calculation formula is as follows:

wherein ,

for the z-th transmission line->

For the probability of failure of the corresponding line +.>

Self-adjusting weather trust factors for corresponding lines.

Further, the method for calculating the self-adjusting expected trust factor and the self-adjusting alarm trust factor comprises the following sub-steps:

a1, constructing a logic state combination among an expected state, a real state and an observed state and corresponding evaluation;

a2, based on the constructed logic state combination and the corresponding evaluation, constructing a logic relation between an expected state and an observed state of the protection device when the action of the protection device is unreliable or the protection alarm information is inaccurate:

similarly, when the action of the circuit breaker is unreliable or alarm information is inaccurate, a logic relationship between the expected state and the observed state of the circuit breaker is constructed:

in the formula,α_r Uncertain finger for protecting deviceStandard, when alpha _r When=1, the protection device is not reliable in operation or the alarm information is not accurate, when α _r When the number is=0, the protection device acts reliably or protects the alarm information accurately; alpha _c As an uncertain index of the circuit breaker, when alpha _c When=1, the circuit breaker is unreliable in action or the alarm information of the circuit breaker is inaccurate, when alpha _c When the circuit breaker is=0, the circuit breaker is reliable in action or the circuit breaker protection alarm information is accurate;

a3, introducing self-checking information to distinguish the uncertain indexes to obtain self-checking alarm indexes s of the protection device _r And a self-checking alarm index s of the circuit breaker _c ；

Wherein, when the protection device sends out self-checking warning, s _r =1, otherwise s _r =0; s when the breaker gives out self-checking alarm _c =1, otherwise s _c ＝0；

A4, combining the uncertain index with the self-checking alarm index to establish a self-adjusting expected trust factor delta of the protection device _rex And self-adjusting alarm trust factor delta _rex ；

δ _rex ＝1-μα _r (1-s _r )

γ _ral ＝1-α _r s _r

Similarly, a self-adjusting expected trust factor delta of the circuit breaker is obtained _cex And self-adjusting alarm trust factor gamma _cal ；

δ _cex ＝1-μα _c (1-s _c )

γ _cal ＝1-α _c s _c

Wherein μ is a desired adjustment coefficient;

the calculation method of the self-adjusting weather trust factor comprises the following steps of:

b1, dividing the line fault risk into four grades according to the external environment condition;

b2, expressing fuzzy relations between line fault risks of various grades and meteorological factors in various disaster weathers by utilizing a triangle membership function, and establishing a gray fuzzy judgment matrix

And the weight matrix of each meteorological factor +.>

And comprehensively judging the risk of the line faults to obtain the risk degree of the line faults>

B3, equally dividing the line fault probability range into 4 sections, and taking the midpoint of each section as the line fault risk degree in one-to-one correspondence with each line fault risk level

Corresponding failure probability D _p ；

B4, calculating a function E according to the alarm information in the SCADA _tl Selected index t of (X) _cp ：

in the formula,

and />

Alarm information link or logic result of main protection, primary backup protection and secondary backup protection of the line head and tail end respectively, < ->

and />

The circuit breaker alarm information is the circuit head end and the circuit breaker alarm information is the circuit tail end respectively, and beta is the minimum coefficient of the alarm information;

b5 based onProbability of failure D _p And selecting index t _cp Calculating a self-adjusting weather trust factor Tp;

T _p ＝2t _cp D _p 。

further, the expected states of the protection device include an action expected of a primary protection, an action expected of a primary backup protection, an action expected of a secondary backup protection, and an action expected of a breaker failure protection;

the action for the main protection expects:

let r be _km Element d being suspected of failing _i If d _i Failure, r _km In response, master protection r _km Motion expectations

The method comprises the following steps:

the actions for the primary backup protection expect:

let r be _kp Element d being suspected of failing _i Primary backup protection of (d) _i Failure and its main protection r _km Refusing movement, r _kp In response, the primary backup protection r _kp Is expected to act

The method comprises the following steps:

the actions for the secondary backup protection expect:

let r be _ks Element d being suspected of failing _i Secondary backup protection of (d), when d _i When the fault occurs, the main protection and the primary backup protection are refused to operate, r _ks Responding; at the same time, when d _x ∈D(R _ks ) If r in case of failure _ks To d _x If none of the circuit breakers on the associated path is operated, r _ks In response to this, the response is that,at this time, the secondary backup protection r _ks Is expected to act

The method comprises the following steps:

in the formula,D(R_ks ) For the collection of network elements within the protection range except for suspected fault elements, a secondary backup protection r _ks The element directly protected is d _i ，p(r _ks ,d _x ) Is r _ks To d _x All breaker sets on the associated path of d _x ∈D；

The action for breaker failure protection is expected to be:

let r be _f For breaker failure protection, when the protection device r _h ∈R(c _h ) Action-driven circuit breaker r _h ∈R(c _h ) C when tripping _h Refusing action, and then protecting the breaker from failure _f In response, the breaker fails to protect r _f Is expected to act

The method comprises the following steps:

in the formula,R(c_h ) To be a drivable circuit breaker c _h Is a set of all protection devices;

the expected state of the circuit breaker is the expected action of the circuit breaker;

let R (c) _k ) To drive the circuit breaker c _k Trip protection device set, when any one can drive c _k Trip protection device r _i When in operation, the circuit breaker c _k In response, the circuit breaker c _k Is expected to act

The method comprises the following steps:

further, the basic unit in the random self-adjusting impulse nerve P system n in the step S4 is an extended impulse nerve P system, and the expression of the extended impulse nerve P system n is as follows:

Π＝(O,σ ₁ ,...,σ _M+2 ,syn,out)

In the formula, the extended impulse nerve P system II is called ESNPS for short;

o= { a } is a single-letter set, a is a nerve pulse, and O is a set of nerve pulses a;

σ ₁ ,...,σ _M+2 for M+2 neurons in a random self-regulating impulse nerve P system, the subscript is 1-M, M is the neuron ordinal number, and neuron sigma _M+1 and σ_M+2 Providing pulses to the system in the same form as the function in the form of sigma _M+1 ＝σ _M+2 = (1, { a→a }) neuron σ _m Expressed in sigma _m ＝(1,R _m ,P _m), wherein ,

for rule set, ++>

As ignition rule when the pulse quantity is varied, +.>

In order to act oppositely, the forgetting rule is expressed in the form of +.>

And

lambda is an empty character generated after execution of the forgetting rule; />

Selecting a limited set of probabilities for the rules, which function to choose the probabilities +.>

and />

Rule(s) respectively corresponding to>

and />

And->

Satisfy the following requirements

syn= { (M, n) ((1.ltoreq.m.ltoreq.m+1)/(n=m+2)) -v ((m=m+2)/(n=m+1)) } is neuron σ _m The connection relation between the two;

out＝{σ ₁ ,…,σ _M the system pi is calculated by neuron sigma and is the output neuron set _m Outputting the calculation result in the form of pulse strings;

the random self-regulating impulse nerve P system comprises two plates, namely an impulse supplier and a binary impulse generator, wherein the impulse supplier comprises a neuron sigma _M+1 and σ_M+2 Generating and delivering pulses to each other, each time the iteration is performed, the pulses are supplied to a binary pulse generator so that each neuron σ _m Can receive a pulse;

the binary pulse generator consists of a neuron sigma ₁ ,...,σ _M The composition and the operation rule are as follows:

(1) During system pi operation, each neuron sigma _m Acts in parallel;

(2) When neuron sigma _m When acting through ignition rules, corresponding neuron sigma _m Outputting binary code 1, otherwise, executing forgetting rule, corresponding to neuron sigma _m Outputting binary code 0;

random self-regulating impulse nerve P systemIn operation, each neuron σ in each ESNPS _m Is set to an ignition rule excitation probability matrix P _R In (b) by controlling matrix P _R The element change in the system realizes the centralized control of all ESNPS;

the ignition rule excitation probability matrix P _R The method comprises the following steps:

in the formula,

for randomly self-regulating the firing rule probability of the ith neuron in the ith expanded impulse nerve P system in the impulse nerve P system, the subscript H is the number of the expanded impulse nerve P systems, and M is the length of the expanded impulse nerve P systems.

Further, in the step S4, the optimizing and solving by using the random self-adjusting pulse neural P system includes the following sub-steps:

S41, using SCADA data as input data of a system II, setting the number of ESNPS in the system as H, the length of each ESNPS as M, and the maximum iteration number as N _maxgen Binary pulse train T _s ；

S42, setting an initial iteration number t=1;

s43, initializing a fitness function according to input data;

s44, rearranging the binary pulse string T _s And converts it into an ignition rule excitation probability matrix P _R ；

Wherein, ignition rule excitation probability matrix P _R Each row corresponds to one ESNPS, which is a neuron sigma in turn ₁ ,...,σ _M Ignition rule excitation probability of (2);

s45, judging whether the current iteration number is smaller than or equal to the maximum iteration number;

if yes, go to step S414;

if not, go to step S46;

s46, exciting the probability matrix P according to the ignition rule _R Generating a solution matrix B so as to meet the following conditions;

B _H×M ＝f _rand <P _R

in the formula,B_H×M Is a solution matrix with the size of H multiplied by M, f _rand Random numbers generated for the rand function;

s47, for each ESNPS, calculating a corresponding fitness function value according to the fitness function fitfunction;

s48, taking the minimum value in the fitness function values corresponding to all ESNPSs as an optimal solution B under the current iteration times _best Corresponding fitness function value f _Bbest ；

At the same time, the maximum value is taken as the worst solution B under the current iteration number _bad Corresponding fitness function value

S49, the fitness function value of the ith individual based on the current iteration times

Updating individual historical optimal solutions

wherein ,

a historical optimal solution for the ith ESNPS;

s410, optimal solution B based on current iteration times _best Updating global optimal solution G _best And its corresponding fitness function value

Meanwhile, based on worst solution B under current iteration times _bad Updating global worst solution G _bad And its corresponding fitness function value

wherein ,G_best I.e. the optimal individual in the previous T times of iteration, G _bad The worst individual in the previous T times of iteration;

s411 randomly generating a learning probability value using rand function

And f is set _rand And->

Comparing the sizes;

s412, according to f _rand And (3) with

Selecting a corresponding learning rate calculation mode to perform ignition rule probability matrix P _R Is updated by learning;

s413, increasing the value of the iteration number T by 1, and returning to the step S45;

s414, the global individual optimal solution G corresponding to the current fitness function value _best As the optimal solution when the random self-adjusting impulse nerve P system is optimized and solved.

Further, in the step S412, when f _rand Less than

When igniting the regular probability matrix P _R The method for learning and updating comprises the following steps:

R1, randomly selecting two individuals k different from the current individual i from H solutions generated by H ESNPSs ₁ and k₂ Determining individual k ₁ and k₂ Whether the corresponding fitness function value satisfies

If yes, entering a step R2;

if not, entering a step R3;

r2, let current individual i go to individual k ₁ Learning, i.e.

Entering a step R4;

r3, let current individual i go to individual k ₂ Learning, i.e.

Entering a step R4;

wherein ,

and />

Is chromosome, is->

Is an intermediate variable +.>

and />

Respectively the kth ₁ 、k ₂ The j-th binary code of the stripe chromosome;

r4, judging the current

Whether or not to establish;

if yes, the current probability ignition rule excites the probability value

The value of increase is +.>

Realizing ignition rule probability matrix P _R Is updated by learning;

if not, the current probability ignition rule excites the probability value

The value of (2) is reduced to->

Realizing ignition rule probability matrix P _R Is updated by learning;

wherein the ignition rule excites probability values

An increase delta of (a) ₁ The calculation formula of (2) is as follows:

ignition rule excitation probability value

Delta of the decrease in (a) ₀ The calculation formula of (2) is as follows:

when f _rand Greater than

When igniting the regular probability matrix P _R The learning updating method comprises the following steps:

judging G under the current iteration number _{best j} Whether or not=1 is true;

if yes, the current ignition rule excites the probability value

Increase to->

Realizing ignition rule probability matrix P _R Is updated by learning;

if not, the current ignition rule excites the probability value

Reduced to->

Realizing ignition rule probability matrix P _R Is updated by learning;

wherein the ignition rule excites probability values

An increment of delta' ₁ The calculation formula of (2) is as follows:

ignition rule excitation probability value

Delta 'of decrease in (2)' ₀ The calculation formula of (2) is as follows:

in the formula,G_{best j} And (3) the j-th bit binary code of the global optimal solution searched under the current iteration times.

Further, in the step S5, when the code of the optimal solution is 0, the corresponding suspected fault element is not faulty; when the code of the optimal solution is 1, the corresponding suspected fault element fails.

The beneficial effects of the invention are as follows:

(1) Because the disaster-causing weather can lead to inaccurate fault alarm information and reduced action reliability of the protection device, trust factors in the fault diagnosis objective function need to be adjusted according to actual conditions;

(2) The invention provides a method for improving the global convergence effect of the fault diagnosis process by utilizing the RSSNPS so as to ensure that the optimal solution is accurately searched, and effectively solve the problems of fault alarm information distortion caused by disaster weather and fault diagnosis under the abnormality of a protection device and a circuit breaker.

Drawings

Fig. 1 is a flowchart of a power grid fault diagnosis method based on a random self-regulating pulse nerve P system.

Fig. 2 is a schematic diagram of an extended impulse nerve P system structure provided by the present invention.

Fig. 3 is a diagram of a random self-regulating impulse nerve P according to the present invention.

Fig. 4 is a network topology diagram in example 1 provided by the present invention.

Fig. 5 is a network topology diagram in example 2 provided by the present invention.

Detailed Description

The following description of the embodiments of the present invention is provided to facilitate understanding of the present invention by those skilled in the art, but it should be understood that the present invention is not limited to the scope of the embodiments, and all the inventions which make use of the inventive concept are protected by the spirit and scope of the present invention as defined and defined in the appended claims to those skilled in the art.

Example 1:

as shown in fig. 1, the power grid fault diagnosis method based on the random self-regulating impulse nerve P system comprises the following steps:

s1, SCADA data is read;

s2, determining a power failure area in the power system to be diagnosed based on the read SCADA data;

S3, establishing a corresponding fault diagnosis objective function based on the determined power failure area;

s4, based on SCADA data, utilizing a random self-adjusting pulse neural P system to perform optimizing solution on the established fault diagnosis objective function to obtain an optimal solution;

s5, determining a power grid fault diagnosis result according to the code of the optimal solution.

In this embodiment, the SCADA data in step S1 is the data recorded with the power grid alarm message and the switching value of the protection device.

In this embodiment, the method for determining the power outage area in step S2 specifically includes:

based on SCADA data, performing iterative search on the electric power system to be diagnosed by using a junction line analysis method, and finding out all passive networks in the electric power system to be diagnosed in the iterative search process, namely determining a power failure area;

the method specifically comprises the following sub-steps:

s21, setting an initial value of the search iteration number C to be 1;

s22, numbering each element in the power system to be diagnosed in sequence based on SCADA data, wherein all element numbers form Q _C ；

S23, slave element number set Q _C Any element number is taken and put into the element number subset M _C In (a) and (b);

s24, judging the latest added element number subset M _C Whether the element corresponding to the element number is provided with a closed circuit breaker connected with the element;

If yes, go to step S24;

if not, go to step S25;

s24, assembling the element numbers into a set Q _C Element numbers corresponding to all elements connected with the currently determined closed circuit breaker are added to the element number subset M _C In the step A3, returning to the step;

s25, increasing the search iteration number C by 1;

s26, slave element number set Q _C-1 Removing element number subset M _C-1 All the element numbers in the database are obtained to obtain a new element number set Q _C ；

S27, judging the current element number set Q _C Whether it is empty;

if yes, go to step S28;

if not, returning to the step S23;

s28, list element number subset M ₁ ,...S _C ,...,M _N Determining a power failure area;

the subscript N is the number of element coding subsets obtained in the iterative search process.

The step S3 of this embodiment specifically includes:

s31, establishing a fault hypothesis H consisting of suspicious fault elements, protection devices and circuit breakers in the power failure area according to the topological structure of the power failure area;

s32, constructing a fault diagnosis objective function based on the fault hypothesis.

The method for establishing the fault hypothesis H in the step S31 specifically includes:

the power failure setting region contains n _d Suspected faulty elements, n _r Protection device and n _c And (3) a breaker, and establishing a fault hypothesis as follows:

X＝[D、R、C]

in the formula,

1≤i≤n _d ，d _i for the actual status of the ith suspect element in the outage area, when d _i At 1, the corresponding suspected failed element has failed, d _i When the value is 0, the corresponding suspicious element is normal and has no fault;

1≤j≤n _r ，r _j for the actual state of the jth protection device (main protection, primary backup protection and secondary backup protection) in the power failure area, when r _j When the value is 1, the corresponding protection device is operated, r _j When the value is 0, the corresponding protection device does not act;

1≤k≤n _c ，c _k when c is the actual state of the kth breaker in the power failure area _k When 1, the corresponding breaker is tripped, c _k And 0, the corresponding circuit breaker is not tripped.

In the step S32, in the disaster-causing weather, the reliability of the actions of some protection devices and the reliability of the information transmitted by the communication system are affected, so that the related protection devices and circuit breakers are more likely to generate the phenomena of refusal and malfunction, and are also more likely to cause the missing report and false report of the alarm information; therefore, a fault diagnosis objective function is established, the conditions of various uncertain factors in the fault diagnosis of the power grid are completely described, the credibility of fault hypothesis is reflected, and the expression of the fault diagnosis objective function minE (X) in the embodiment is as follows:

min E(X)＝δ _ex E _ex (X)+γ _al E _al (X)+T _p E _tl (X)

in the formula,δ_ex To self-regulate desired trust factor, self-regulate desired trust factor comprising protection means

And self-adjusting desired trust factor of the circuit breaker>

γ _al For self-regulating alarm trust factor, self-regulating alarm trust factor gamma comprising protection means _ral Self-adjusting alarm trust factor gamma with circuit breaker _cal ；

T _p Is a self-adjusting weather trust factor;

E _ex (X) is a function reflecting the malfunction and refusal conditions of the protection device and the circuit breaker, the smaller the function value is, the smaller the number of the malfunction protection device and the refusal protection device is, the smaller the number of the surface is, and the calculation formula is as follows:

wherein ,

self-adjusting desired trust factor for the j-th protection means>

Self-adjusting desired trust factor for the kth circuit breaker correspondence,>

for the desired state of the jth protective device, < >>

Expectations for the kth circuit breakerA state;

E _al the function value of the function is the quantity of the information of the protection device which is not reported and is reported by mistake, the smaller the value is, the smaller the indication quantity is, and the calculation formula is as follows:

wherein ,

self-adjusting alarm trust factor for the j-th protection device,>

self-adjusting alarm trust factor corresponding to kth circuit breaker, >

For the observation state of the jth protective device, < >>

The observation state of the kth circuit breaker;

E _tl (X) is a function reflecting the matching condition of the fault probability and the real state of the line, the fault caused by the extreme weather is judged by using weather information, misdiagnosis caused by the influence of external environment is avoided, the smaller the value is, the fault probability of the line is approximately matched with the real state, and the calculation formula is as follows:

wherein ,

for the z-th transmission line->

For the probability of failure of the corresponding line +.>

Self-adjusting weather trust factors for corresponding lines.

Specifically, in practical application, for the same protection device or breaker, the following two cases are both small probability events, namely false action occurs and alarm information is missed, or false alarm occurs when action rejection occurs, so in order to reduce calculation complexity, the calculation of the self-adjustment expected trust factor and the self-adjustment alarm trust factor comprises the following sub-steps:

a1, constructing a logic state combination among an expected state, a real state and an observed state and corresponding evaluation; (Table 1)

Table 1 logical state combinations and corresponding evaluations

r ex	r	r al	Judging
				0	0	0	Accurate and accurate
0	0	1	False alarm
				0	1	0	Small probability
0	1	1	Malfunction of the switch
				1	0	0	Refusing movement
1	0	1	Small probability
				1	1	0	Missing report
1	1	1	Accurate and accurate

A2, based on the constructed logic state combination and the corresponding evaluation (table 1), when the action of the protection device is unreliable or the protection alarm information is inaccurate, constructing a logic relation between the expected state and the observed state of the protection device:

similarly, when the action of the circuit breaker is unreliable or alarm information is inaccurate, a logic relationship between the expected state and the observed state of the circuit breaker is constructed:

in the formula,α_r As an uncertainty index of the protection device, when alpha _r When=1, the protection device is not reliable in operation or the alarm information is not accurate, when α _r When the number is=0, the protection device acts reliably or protects the alarm information accurately; when alpha is _c As an uncertain index of the circuit breaker, when alpha _c When=1, the circuit breaker is unreliable in action or the alarm information of the circuit breaker is inaccurate, when alpha _c When the circuit breaker is=0, the circuit breaker is reliable in action or the circuit breaker protection alarm information is accurate;

a3, introducing self-checking information to distinguish the uncertain indexes to obtain self-checking alarm indexes s of the protection device _r And a self-checking alarm index s of the circuit breaker _c ；

Wherein, when the protection device sends out self-checking warning, s _r =1, otherwise s _r =0; s when the breaker gives out self-checking alarm _c =1, otherwise s _c ＝0；

Specifically, self-checking information is introduced to judge uncertain indexes, and the two conditions of unreliable protection devices or inaccurate alarm information are distinguished, so that operation shows that when the state of the protection devices is abnormal, the protection devices send out alarms through the self-checking technology, self-checking information alarms when the conditions of missing and false alarms of alarm information are assumed, but the missing and false alarms do not necessarily occur when the self-checking alarms occur;

A4, combining the uncertain index with the self-checking alarm index to establish a self-adjusting expected trust factor delta of the protection device _rex And self-adjusting alarm trust factor delta _rex ；

δ _rex ＝1-μα _r (1-s _r )

γ _ral ＝1-α _r s _r

Similarly, a self-adjusting expected trust factor delta of the circuit breaker is obtained _cex And self-adjusting alarmsTrust factor gamma _cal ；

δ _cex ＝1-μα _c (1-s _c )

γ _cal ＝1-α _c s _c

Where μ is a desired adjustment coefficient, and is set by the user according to circumstances, and is set to 0.3 in this embodiment.

Specifically, the method for calculating the self-adjusting weather trust factor comprises the following sub-steps:

b1, dividing the line fault risk into four grades according to the external environment condition; wherein the four grades are high risk, higher risk, lower risk and low risk, respectively;

b2, expressing fuzzy relations between line fault risks of various grades and meteorological factors in various disaster weathers by utilizing a triangle membership function, and establishing a gray fuzzy judgment matrix

And the weight matrix of each meteorological factor +.>

And comprehensively judging the risk of the line faults to obtain the risk degree of the line faults>

B3, equally dividing the line fault probability range into 4 sections, and taking the midpoint of each section as the line fault risk degree in one-to-one correspondence with each line fault risk level

Corresponding failure probability D _p ；

The fault probability corresponding to the line fault risk is shown in table 2:

Table 2 probability of failure corresponding to the line failure risk

Risk of line faults	Corresponding failure probability range	Corresponding probability of failure
			High height	0.75～1.00	0.875
Higher height	0.50～0.75	0.625
			Lower level	0.25～0.50	0.375
Low and low	0.00～0.25	0.125

B4, calculating a function E according to the alarm information in the SCADA _tl Selected index t of (X) _cp ：

/>

in the formula,

and />

The alarm information is the logical result of the primary protection, the primary backup protection and the secondary backup protection of the head end and the tail end of the line respectively. />

and />

And the circuit breaker alarm information is respectively the head end and the tail end of the circuit. Beta is the minimum coefficient of alarm information, which means the reciprocal of the maximum value of the alarm information per power line in the case of failure, and is set to 0.25 in this embodiment. When t _cp When the numerical value of (2) is more than or equal to 0.5, taking 1; t is t _cp When the value is less than 0.5, taking 0;

b5 based on failure probability D _p And selecting index t _cp Calculating a self-adjusting weather trust factor Tp;

T _p ＝2t _cp D _p

the influence degree of the weather information of the main condition of the weather trust factor on the fault diagnosis objective function is self-regulated, and misjudgment of the objective function on the condition that the high risk of the line is not faulty and the low risk of the line is faulty is prevented.

The expected states of the protection device comprise the expected action of the main protection, the expected action of the primary backup protection and the expected action of the failure protection of the breaker;

The action for the main protection expects:

let r be _km Element d being suspected of failing _i If d _i Failure, r _km In response, master protection r _km Motion expectations

The method comprises the following steps:

the actions for the primary backup protection expect:

let r be _kp Element d being suspected of failing _i Primary backup protection of (d) _i Failure and its main protection r _km Refusing movement, r _kp In response, the primary backup protection r _kp Is expected to act

The method comprises the following steps:

the actions for the secondary backup protection expect:

let r be _ks Element d being suspected of failing _i Secondary backup protection of (d), when d _i When the fault occurs, the main protection and the primary backup protection are refused to operate, r _ks Responding; at the same time, when d _x ∈D(R _ks ) If r in case of failure _ks To d _x If none of the circuit breakers on the associated path is operated, r _ks In response, the secondary backup protection r _ks Is expected to act

The method comprises the following steps:

in the formula,D(R_ks ) For the collection of network elements within the protection range except for suspected fault elements, a secondary backup protection r _ks The element directly protected is d _i ，p(r _ks ,d _x ) Is r _ks To d _x All breaker sets on the associated path of d _x ∈D；

The action for breaker failure protection is expected to be:

let r be _f For breaker failure protection, when the protection device r _h ∈R(c _h ) Action-driven circuit breaker r _h ∈R(c _h ) C when tripping _h Refusing action, and then protecting the breaker from failure _f In response, the breaker fails to protect r _f Is expected to act

The method comprises the following steps:

in the formula,R(c_h ) To be a drivable circuit breaker c _h Is a set of all protection devices;

the expected state of the circuit breaker is the expected action of the circuit breaker;

let R (c) _k ) To drive the circuit breaker c _k Trip protection device set, when any one can drive c _k Trip protection device r _i When in operation, the circuit breaker c _k In response, the circuit breaker c _k Is expected to act

The method comprises the following steps: />

The basic unit in the random self-regulating impulse nerve P system (RSSNPS) in step S4 of the present embodiment is an extended impulse nerve P system (Extend Spiking Neural P Systems, ESNPS), and the expression of the extended impulse nerve P system pi is:

Π＝(O,σ ₁ ,...,σ _M+2 ,syn,out)

wherein O= { a } is a single-letter set, a is a nerve pulse, and O is a set of nerve pulses a;

σ ₁ ,...,σ _M+2 for M+2 neurons in a random self-regulating impulse nerve P system, the subscript is 1-M, M is the neuron ordinal number, and neuron sigma _M+1 and σ_M+2 Providing pulses to the system in the same form as the function in the form of sigma _M+1 ＝σ _M+2 = (1, { a→a }) neuron σ _m Expressed in sigma _m ＝(1,R _m ,P _m), wherein ,

for rule set, ++>

As ignition rule when the pulse quantity is varied, +.>

In order to act oppositely, the forgetting rule is expressed in the form of +.>

And

lambda is an empty character generated after execution of the forgetting rule; / >

Selecting a limited set of probabilities for the rules, which function to choose the probabilities +.>

and />

Rule(s) respectively corresponding to>

and />

And->

Satisfy the following requirements

syn= { (M, n) ((1.ltoreq.m.ltoreq.m+1)/(n=m+2)) -v ((m=m+2)/(n=m+1)) } is neuron σ _m Connection relation (synapse) between them;

out＝{σ ₁ ,…,σ _M the system pi is calculated by neuron sigma and is the output neuron set _m In the form of pulse trainsOutputting a calculation result;

as can be seen from fig. 2, the extended impulse nerve P system comprises two plates, respectively an impulse feeder in which the neurons σ, and a binary impulse generator _M+1 and σ_M+2 Generating and delivering pulses to each other, each time the iteration is performed, the pulses are supplied to a binary pulse generator so that each neuron σ _m Can receive a pulse;

binary pulse generator is composed of neuron sigma ₁ ,...,σ _M The composition and the operation rule are as follows:

(1) During system pi operation, each neuron sigma _m Acts in parallel;

(2) When neuron sigma _m When acting through ignition rules, corresponding neuron sigma _m Outputting binary code 1, otherwise, executing forgetting rule, corresponding to neuron sigma _m A binary code 0 is output.

From the structure of system II, the output of the system is a binary pulse train consisting of '1' and '0', and the output result and probability of each neuron

and />

Related, probability->

Control the output of "1", probability->

The output of "0" is controlled. Because of->

and />

There is->

So that (3) is a mathematical relationshipThe key to the control regulation of the whole system is +.>

Is controlled and regulated.

By arranging H ESNPSs (Extended Spiking Neural P Systems, expanded impulse nerve P system) in parallel, and adding probability matrix P _R Control management to enable each neuron σ in each ESNPS _m The regular excitation probability of (m=1, 2, … M) is controlled centrally, and the structure thereof is shown in fig. 3:

specifically, to enable a random self-regulating pulsed neural P system to effectively control all ESNPSs, each neuron σ in each ESNPS is controlled _m Is set to an ignition rule excitation probability matrix P _R By controlling the matrix P _R The element in the ESNPS is changed to achieve the purpose of centralized control of all ESNPS in parallel;

probability matrix P _R The following is shown:

in the formula,

for randomly self-regulating the firing rule probability of the jth neuron in the ith expanded impulse nerve P system in the impulse nerve P system, the subscript H is the number of the expanded impulse nerve P systems, and M is the length of the expanded impulse nerve P systems.

Therefore, the probability matrix P is excited by H parallel ESNPS and an ignition rule _R And the system with randomly selected learning and self-regulating learning rate variation is a randomly self-regulating pulse nerve P system.

Based on the system structure, in the step S4, when the random self-adjusting pulse neural P system is used for optimizing and solving, the method comprises the following sub-steps:

s41, using SCADA data as input data of a system II, setting the number of ESNPS in the system as H, and the length of each ESNPSDegree is M, the maximum iteration number is N _maxgen Binary pulse train T _s ；

S42, setting an initial iteration number t=1;

s43, initializing a fitness function according to input data;

s44, rearranging the binary pulse string T _s And converts it into an ignition rule excitation probability matrix P _R ；

Wherein, ignition rule excitation probability matrix P _R Each row corresponds to one ESNPS, which is a neuron sigma in turn ₁ ,...,σ _M Ignition rule excitation probability of (2);

s45, judging whether the current iteration number is smaller than or equal to the maximum iteration number;

if yes, go to step S414;

if not, go to step S46;

s46, exciting the probability matrix P according to the ignition rule _R Generating a solution matrix B so as to meet the following conditions;

B _H×M ＝f _rand <P _R

in the formula,B_H×M Is a solution matrix with the size of H multiplied by M, f _rand Random numbers generated for the rand function;

S47, for each ESNPS, calculating a corresponding fitness function value according to the fitness function fitfunction;

s48, taking the minimum value in the fitness function values corresponding to all ESNPSs as an optimal solution B under the current iteration times _best Corresponding fitness function value f _Bbest ；

At the same time, the maximum value is taken as the worst solution B under the current iteration number _bad Corresponding fitness function value

S49, the fitness function value of the ith individual based on the current iteration times

Updating individual historical optimal solutions

wherein ,

a historical optimal solution for the ith ESNPS;

s410, optimal solution B based on current iteration times _best Updating global optimal solution G _best And its corresponding fitness function value

Meanwhile, based on worst solution B under current iteration times _bad Updating global worst solution G _bad And its corresponding fitness function value

/>

wherein ,G_best I.e. the optimal individual in the previous T times of iteration, G _bad The worst individual in the previous T times of iteration;

s411 randomly generating a learning probability value using rand function

And f is set _rand And->

Comparing the sizes;

s412, according to f _rand And (3) with

Selecting a corresponding learning rate calculation mode to perform ignition rule probability matrix P _R Is updated by learning;

s413, increasing the value of the iteration number T by 1, and returning to the step S45;

S414, the global individual optimal solution G corresponding to the current fitness function value _best As the optimal solution when the random self-adjusting impulse nerve P system is optimized and solved.

In particularIn step S412, when f _rand Less than

When igniting the regular probability matrix P _R The method for learning and updating comprises the following steps:

r1, randomly selecting two individuals k different from the current individual i from H solutions generated by H ESNPSs ₁ and k₂ Determining individual k ₁ and k₂ Whether the corresponding fitness function value satisfies

If yes, entering a step R2;

if not, entering a step R3;

r2, let current individual i go to individual k ₁ Learning, i.e.

Entering a step R4;

r3, let current individual i go to individual k ₂ Learning, i.e.

Entering a step R4;

wherein ,

and />

Is chromosome, is->

Is an intermediate variable +.>

and />

Respectively the kth ₁ 、k ₂ The j-th binary code of the stripe chromosome;

r4, judging the current

Whether or not to establish;

if yes, the current probability ignition rule excites the probability value

The value of increase is +.>

Realizing ignition rule probability matrix P _R Is updated by learning;

if not, the current probability ignition rule excites the probability value

The value of (2) is reduced to->

Realizing ignition rule probability matrix P _R Is updated by learning;

wherein the ignition rule excites probability values

An increase delta of (a) ₁ The calculation formula of (2) is as follows:

ignition rule excitation probability value

Delta of the decrease in (a) ₀ The calculation formula of (2) is as follows:

when f _rand Greater than

When igniting the regular probability matrix P _R The learning updating method comprises the following steps:

judging G under the current iteration number _{best j} Whether or not=1 is true;

if yes, the current ignition rule excites the probability value

Increase to->

Realizing ignition rule probability matrix P _R Is updated by learning;

if not, the current ignition rule excites the probability value

Reduced to->

Realizing ignition rule probability matrix P _R Is updated by learning;

wherein the ignition rule excites probability values

An increment of delta' ₁ The calculation formula of (2) is as follows:

ignition rule excitation probability value

Delta 'of decrease in (2)' ₀ The calculation formula of (2) is as follows:

in the formula,G_{best j} And (3) the j-th bit binary code of the global optimal solution searched under the current iteration times.

In step S5 of this embodiment, when the code of the optimal solution is 0, the corresponding suspected fault element is not faulty; when the code of the optimal solution is 1, the corresponding suspicious fault element fails.

Example 2:

in this embodiment, the following two examples are taken as examples, and a specific power grid fault diagnosis process based on the method is given.

The fault scenario for example 1 is as follows:

there are 28 elements, 40 circuit breakers, 124 protections and 40 breaker failure protections in the system. Here, A, B denotes a busbar, T denotes a transformer, L denotes a line, S, R denotes the first and the second ends of the line (the first and the second ends of the line are defined from top to bottom and from left to right), km denotes a main protection, kp denotes a first backup protection, ks denotes a second backup protection, and f denotes a breaker failure protection.

The transformer T3 and the bus B2 simultaneously fail, and the protection and circuit breaker act as follows: the transformer main protection T3m acts to trip QF16, QF14 refuses to act, the failure protection QF14f acts to trip QF12, QF13 and QF19; the bus B2 mainly protects B2m to act, and breaks away QF4, QF6, QF8 and QF10, and the line protection L3Rks malfunctions, and QF27 trips, and alarm information of actions of T3m, T3p, L3Rs, QF14f, QF4, QF6, QF8, QF10, QF12, QF16, QF19 and QF27 is received.

The blackout area is shown in fig. 4 in darkened form, and in case of diagnostic example 1, it is assumed that lines L2 and L3 are not damaged, at lower risk,

all take 0.375, construct the fault hypothesis in the specific form:

X＝[d ₁ ,...,d ₅ ,r ₁ ,...,r ₂₃ ,c ₁ ,...,c ₁₀ ]

the blackout area is shown in fig. 4 in darkened form, and in case of diagnostic example 1, it is assumed that lines L2 and L3 are not damaged, at lower risk,

all taken at 0.375. And according to the self-checking information, B2m self-checking alarms. The specific form of constructing the fault hypothesis is as follows:

X＝[d ₁ ,...,d ₅ ,r ₁ ,...,r ₂₃ ,c ₁ ,...,c ₁₀ ]

the related device sets involved are as follows:

suspicious faulty element set: there are 5 elements in the outage area: b2, B4, T3, L2, L3 are all suspected fault elements, then d= { d ₁ ,d ₂ ,d ₃ ,d ₄ ,d ₅ }。

A set of related circuit breakers: involving 10 circuit breakers in total, namely QF4, QF6, QF8, QF10, QF12, QF13, QF14, QF16, QF19, QF27, thus C= { C ₁ ,c ₂ ,...,c ₁₀ }。

Related protection equipment: a total of 23 protections: t3km, T3kp, T3ks, B4m, B2m, L2Skm, L2Rkm, L2Skp, L2Rkp, L2Sks, L2Rks, L3Skm, L3Skp, L3Rkp, L3Sks, L3Rks, QF4f, QF8f, QF10f, QF12f, QF14f, QF19f, thus R= { R ₁ ,r ₂ ,...,r ₂₃ }。

From the received alarm information, it can be determined that:

based on the received self-test information, it can be determined that the self-test alarm index s _r5 1 and the balance 0. The fault scenario for example 2 is as follows:

in the system, B represents a bus bar, L represents a line, S, R represents the first and the second ends of the line (the first and the second ends of the line are defined according to the sequence of the line, which are sequentially connected), km represents a main protection, kp represents a first backup protection, ks represents a second backup protection, and f represents a breaker failure protection.

Fault background conditions: the faults of L1 and L3 are caused by strong wind, the main protection actions of the circuits L1 and L3 are performed, and the circuit breakers C1, C2, C4 and C6 are tripped; the bus B1 protects the operation and trips the circuit breakers C4 and C7. In the fault process, the information distortion phenomenon of fault alarm information occurs in the transmission process due to lightning interference. The result is that the protection of the line L3 approaching the Rainbow change and the protection information of the line L1 approaching the wide A outgoing line are not reported, the action information of the circuit breakers C1 and C6 are not reported, and the action information of the circuit breaker C3 is misreported.

The blackout area is used in FIG. 5Shadow marking, assuming a disaster period, line L ₁ 、L ₃ At the risk of being at a high level,

take 0.875, L ₂ At lower risk,/->

Taken as 0.375. From the self-test information, L1Skm, L3Rkm, QF1, QF3, QF6 self-test alarms are known. Since no 1 case breaker failure protection alarm occurs, no breaker failure protection is involved in the hypothesis. The specific form of constructing the fault hypothesis is as follows:

X＝[d ₁ ,...,d ₅ ,r ₁ ,...,r ₂₀ ,c ₁ ,...,c ₇ ]

the related device sets involved are as follows:

suspicious faulty element set: there are 5 elements in the outage area: l1, L2, L3, B1, B3 are all suspected fault elements, thus d= { d ₁ ,d ₂ ,d ₃ ,d ₄ ,d ₅ }。

A set of related circuit breakers: involving 7 circuit breakers in total, namely QF1, QF2, QF3, QF4, QF5, QF6, QF7, thus C= { C ₁ ,c ₂ ,…,c ₇ }。

Related protection equipment: a total of 20 protections: l1Skm, L1Skp, L1Sks, L1Rkm, L1Rkp, L1Rks, L2Skm, L2Skp, L2Sks, L2Rkm, L2Rkp, L2Rks, L3Skm, L3Skp, L3Sks, L3Rkm, L3Rkp, L3Rks, B1m, B3m, then R= { R ₁ ,r ₂ ,…,r ₂₀ }。

From the received alarm information, it can be determined that:

based on the received self-test information, it can be determined that the self-test alarm index s _r1 、s _r16 、s _c1 、s _c3 、s _c6 1 and the balance 0. Substituting the fault scene information of each of the calculation examples 1 and 2 into each of the otherIn the fault diagnosis model, diagnosis is carried out according to a diagnosis flow, the fault diagnosis model established by two fault systems is solved by adopting RSSNPS, the hypothesis X= (D, R, C) is used as a population individual, and the coding lengths are 38 and 32 bits respectively. The population sizes were 50 and the maximum number of iterations was 150. Example 1 took about 3 seconds and example 2 took about 2.4 seconds. The diagnosis results are as follows:

TABLE 3 diagnosis results of EXAMPLE 1

TABLE 4 diagnosis results of example 2

The diagnosis result shows that the invention can accurately diagnose the fault element when the fault element is in face of malfunction, refusal of the protection device and the alarm information is distorted. When the RSSNPS is used for calculating the established objective function, the overall optimizing capability is strong, the convergence speed is high, and the convergence effect is good.

Claims (1)

1. The power grid fault diagnosis method based on the random self-regulating pulse nerve P system is characterized by comprising the following steps of:

s1, SCADA data is read;

s2, determining a power failure area in the power system to be diagnosed based on the read SCADA data;

s3, establishing a corresponding fault diagnosis objective function based on the determined power failure area;

s4, based on SCADA data, utilizing a random self-adjusting pulse neural P system to perform optimizing solution on the established fault diagnosis objective function to obtain an optimal solution;

s5, determining a power grid fault diagnosis result according to the code of the optimal solution;

the method for determining the power failure area in the step S2 specifically includes:

based on SCADA data, performing iterative search on the electric power system to be diagnosed by using a junction line analysis method, and finding out all passive networks in the electric power system to be diagnosed in the iterative search process, namely determining a power failure area;

The method specifically comprises the following sub-steps:

s21, setting an initial value of the search iteration number C to be 1;

s22, numbering each element in the power system to be diagnosed in sequence based on SCADA data, wherein all element numbers form Q _C ；

S23, slave element number set Q _C Any element number is taken and put into the element number subset M _C In (a) and (b);

s24, judging the latest added element number subset M _C Whether the element corresponding to the element number is provided with a closed circuit breaker connected with the element;

if yes, go to step S24;

if not, go to step S25;

s24, assembling the element numbers into a set Q _C Element numbers corresponding to all elements connected with the currently determined closed circuit breaker are added to the element number subset M _C In the step A3, returning to the step;

s25, increasing the search iteration number C by 1;

s26, slave element number set Q _C-1 Removing element number subset M _C-1 All the element numbers in the database are obtained to obtain a new element number set Q _C ；

S27, judging the current element number set Q _C Whether it is empty;

if yes, go to step S28;

if not, returning to the step S23;

s28, list element number subset M ₁ ,...S _C ,...,M _N Determining a power failure area;

the subscript N is the number of element coding subsets obtained in the iterative search process;

The step S3 specifically comprises the following steps:

s31, establishing a fault hypothesis H consisting of suspicious fault elements, protection devices and circuit breakers in the power failure area according to the topological structure of the power failure area;

s32, constructing a fault diagnosis objective function based on a fault hypothesis;

the method for establishing the fault hypothesis H in step S31 specifically includes:

the power failure setting region contains n _d Suspected faulty elements, n _r Protection device and n _c And (3) a breaker, and establishing a fault hypothesis as follows:

X＝[D、R、C]

in the formula,

1≤i≤n _d ，d _i for the actual status of the ith suspect element in the outage area, when d _i At 1, the corresponding suspected failed element has failed, d _i When the value is 0, the corresponding suspicious element is normal and has no fault;

1≤j≤n _r ，r _j for the actual state of the jth protection device in the power failure area, when r _j When the value is 1, the corresponding protection device is operated, r _j When the value is 0, the corresponding protection device does not act;

1≤k≤n _c ，c _k when c is the actual state of the kth breaker in the power failure area _k When 1, the corresponding breaker is tripped, c _k 0, corresponding to the breaker not tripped;

in the step S32, the expression of the fault diagnosis objective function minE (X) is:

minE(X)＝δ _ex E _ex (X)+γ _al E _al (X)+T _p E _tl (X)

in the formula,δ_ex To self-regulate desired trust factor, self-regulate desired trust factor comprising protection means

And self-adjusting desired trust factor of the circuit breaker >

γ _al For self-regulating alarm trust factor, self-regulating alarm trust factor gamma comprising protection means _ral Self-adjusting alarm trust factor gamma with circuit breaker _cal ；

T _p Is a self-adjusting weather trust factor;

E _ex (X) is a function reflecting the malfunction and refusal of the protection device and the circuit breaker, and the calculation formula is as follows:

wherein ,

self-adjusting desired trust factor for the j-th protection means>

Self-adjusting desired trust factor for the kth circuit breaker correspondence,>

for the desired state of the jth protective device, < >>

Is the expected state of the kth circuit breaker;

E _al the (X) is a function reflecting the condition of missing report and false report of alarm information of a protection device and a circuit breaker, and the calculation formula is as follows:

wherein ,

self-adjusting alarm trust factor for the j-th protection device,>

self-adjusting alarm trust factor corresponding to kth circuit breaker,>

for the observation state of the jth protective device, < >>

The observation state of the kth circuit breaker;

E _tl (X) is a function reflecting the matching condition of the fault probability and the real state of the line, and the calculation formula is as follows:

wherein ,

for the z-th transmission line->

For the probability of failure of the corresponding line +.>

Self-adjusting weather trust factors for corresponding lines;

the calculation method of the self-adjusting expected trust factor and the self-adjusting alarm trust factor comprises the following substeps:

A1, constructing a logic state combination among an expected state, a real state and an observed state and corresponding evaluation;

a2, based on the constructed logic state combination and the corresponding evaluation, constructing a logic relation between an expected state and an observed state of the protection device when the action of the protection device is unreliable or the protection alarm information is inaccurate:

similarly, when the action of the circuit breaker is unreliable or alarm information is inaccurate, a logic relationship between the expected state and the observed state of the circuit breaker is constructed:

in the formula,α_r As an uncertainty index of the protection device, when alpha _r When=1, the protection device is not reliable in operation or the alarm information is not accurate, when α _r When the number is=0, the protection device acts reliably or protects the alarm information accurately; alpha _c As an uncertain index of the circuit breaker, when alpha _c When=1, the circuit breaker is unreliable in action or the alarm information of the circuit breaker is inaccurate, when alpha _c When the circuit breaker is=0, the circuit breaker is reliable in action or the circuit breaker protection alarm information is accurate;

a3, introducing self-checking information to distinguish the uncertain indexes to obtain self-checking alarm indexes s of the protection device _r And a self-checking alarm index s of the circuit breaker _c ；

Wherein, when the protection device sends out self-checking warning, s _r =1, otherwise s _r =0; s when the breaker gives out self-checking alarm _c =1, otherwise s _c ＝0；

A4, combining the uncertain index with the self-checking alarm index to establish a self-adjusting expected trust factor delta of the protection device _rex And self-adjusting alarm trust factor delta _rex ；

δ _rex ＝1-μα _r (1-s _r )

γ _ral ＝1-α _r s _r

Similarly, a self-adjusting expected trust factor delta of the circuit breaker is obtained _cex And self-adjusting alarm trust factor gamma _cal ；

δ _cex ＝1-μα _c (1-s _c )

γ _cal ＝1-α _c s _c

Wherein μ is a desired adjustment coefficient;

the calculation method of the self-adjusting weather trust factor comprises the following steps of:

b1, dividing the line fault risk into four grades according to the external environment condition;

b2, expressing fuzzy relations between line fault risks of various grades and meteorological factors in various disaster weathers by utilizing a triangle membership function, and establishing a gray fuzzy judgment matrix

And the weight matrix of each meteorological factor +.>

And comprehensively judging the risk of the line faults to obtain the risk degree of the line faults>

B3, equally dividing the line fault probability range into 4 sections, and taking the midpoint of each section as the line fault risk degree in one-to-one correspondence with each line fault risk level

Corresponding failure probability D _p ；

B4, calculating a function E according to the alarm information in the SCADA _tl Selected index t of (X) _cp ：

in the formula,

and />

Alarm information link or logic result of main protection, primary backup protection and secondary backup protection of the line head and tail end respectively, < - >

and />

The circuit breaker alarm information is the circuit head end and the circuit breaker alarm information is the circuit tail end respectively, and beta is the minimum coefficient of the alarm information;

b5 based on failure probability D _p And selecting index t _cp Calculating a self-adjusting weather trust factor Tp;

T _p ＝2t _cp D _p

the expected states of the protection device comprise an action expected of a main protection, an action expected of a primary backup protection, an action expected of a secondary backup protection and an action expected of a breaker failure protection;

the action for the main protection expects:

let r be _km Element d being suspected of failing _i If d _i Failure, r _km In response, master protection r _km Motion expectations

The method comprises the following steps:

the actions for the primary backup protection expect:

let r be _kp Element d being suspected of failing _i Primary backup protection of (d) _i Failure and its main protection r _km Refusing movement, r _kp In response, the primary backup protection r _kp Is expected to act

The method comprises the following steps:

the actions for the secondary backup protection expect:

let r be _ks Element d being suspected of failing _i Secondary backup protection of (d), when d _i When the fault occurs, the main protection and the primary backup protection are refused to operate, r _ks Responding; at the same time, when d _x ∈D(R _ks ) If r in case of failure _ks To d _x If none of the circuit breakers on the associated path is operated, r _ks In response, the secondary backup protection r _ks Is expected to act

The method comprises the following steps:

in the formula,D(R_ks ) For the collection of network elements within the protection range except for suspected fault elements, a secondary backup protection r _ks The element directly protected is d _i ，p(r _ks ,d _x ) Is r _ks To d _x All breaker sets on the associated path of d _x ∈D；

The action for breaker failure protection is expected to be:

let r be _f For breaker failure protection, when the protection device r _h ∈R(c _h ) Action-driven circuit breaker r _h ∈R(c _h ) C when tripping _h Refusing action, and then protecting the breaker from failure _f In response, the breaker fails to protect r _f Is expected to act

The method comprises the following steps:

in the formula,R(c_h ) To be a drivable circuit breaker c _h Is a set of all protection devices;

the expected state of the circuit breaker is the expected action of the circuit breaker;

let R (c) _k ) To drive the circuit breaker c _k Trip protection device set, when any one can drive c _k Trip protection device r _i When in operation, the circuit breaker c _k In response, the circuit breaker c _k Is expected to act

The method comprises the following steps:

the basic unit in the random self-adjusting impulse nerve P system in the step S4 is an extended impulse nerve P system, and the expression of the extended impulse nerve P system pi is as follows:

Π＝(O,σ ₁ ,...,σ _M+2 ,syn,out)

in the formula, the extended impulse nerve P system II is called ESNPS for short;

o= { a } is a single-letter set, a is a nerve pulse, and O is a set of nerve pulses a;

σ ₁ ,...,σ _M+2 For M+2 neurons in a random self-regulating impulse nerve P system, the subscript is 1-M, M is the neuron ordinal number, and neuron sigma _M+1 and σ_M+2 Providing pulses to the system in the same form as the function in the form of sigma _M+1 ＝σ _M+2 = (1, { a→a }) neuron σ _m Expressed in sigma _m ＝(1,R _m ,P _m), wherein ,

for rule set, ++>

As ignition rule when the pulse quantity is varied, +.>

In order to act oppositely, the forgetting rule is expressed in the form of +.>

And

lambda is an empty character generated after execution of the forgetting rule; />

Selecting a limited set of probabilities for the rules, which function to choose the probabilities +.>

and />

Rule(s) respectively corresponding to>

and />

And->

Satisfy the following requirements

syn= { (M, n) ((1.ltoreq.m.ltoreq.m+1)/(n=m+2)) -v ((m=m+2)/(n=m+1)) } is neuron σ _m The connection relation between the two;

out＝{σ ₁ ,…,σ _M the system pi is calculated by neuron sigma and is the output neuron set _m Outputting the calculation result in the form of pulse strings;

the extended impulse nerve P system comprises two plates, namely an impulse supplier and a binary impulse generator, wherein the impulse supplier comprises a neuron sigma _M+1 and σ_M+2 Generating and delivering pulses to each other, each time the iteration is performed, the pulses are supplied to a binary pulse generator so that each neuron σ _m Can receive a pulse;

the binary pulse generator consists of a neuron sigma ₁ ,...,σ _M The composition and the operation rule are as follows:

(1) During system pi operation, each neuron sigma _m Acts in parallel;

(2) When neuron sigma _m When acting through ignition rules, corresponding neuron sigma _m Outputting binary code 1, otherwise, executing forgetting rule, corresponding to neuron sigma _m Outputting binary code 0;

when the random self-regulating impulse nerve P system is running, each neuron sigma in each ESNPS _m Is set to an ignition rule excitation probability matrix P _R In (b) by controlling matrix P _R The element change in the system realizes the centralized control of all ESNPS;

the ignition rule excitation probability matrix P _R The method comprises the following steps:

in the formula,

for randomly self-regulating the firing rule probability of the jth neuron in the ith expanded impulse nerve P system in the impulse nerve P system, the subscript H is the number of the expanded impulse nerve P systems, and M is the length of the expanded impulse nerve P systems;

in the step S4, the optimizing and solving by using the random self-adjusting pulse neural P system includes the following sub-steps:

s41, using SCADA data as input data of a system II, setting the number of ESNPS in the system as H, the length of each ESNPS as M, and the maximum iteration number as N _maxgen Binary pulse train T _s ；

S42, setting an initial iteration number t=1;

s43, initializing a fitness function according to input data;

s44, rearranging the binary pulse string T _s And converts it into an ignition rule excitation probability matrix P _R ；

Wherein, ignition rule excitation probability matrix P _R Each row corresponds to one ESNPS, which is a neuron sigma in turn ₁ ,...,σ _M Ignition rule excitation probability of (2);

s45, judging whether the current iteration number is smaller than or equal to the maximum iteration number;

if yes, go to step S414;

if not, go to step S46;

s46, exciting the probability matrix P according to the ignition rule _R Generating a solution matrix B so as to meet the following conditions;

B _H×M ＝f _rand <P _R

in the formula,B_H×M Is a solution matrix with the size of H multiplied by M, f _rand Random numbers generated for the rand function;

s47, for each ESNPS, calculating a corresponding fitness function value according to the fitness function fitfunction;

s48, taking the minimum value in the fitness function values corresponding to all ESNPSs as an optimal solution B under the current iteration times _best Corresponding fitness function value f _Bbest ；

At the same time, the maximum value is taken as the worst solution B under the current iteration number _bad Corresponding fitness function value

S49, the fitness function value of the ith individual based on the current iteration times

Updating individual history optimal solution->

wherein ,

a historical optimal solution for the ith ESNPS;

s410, optimal solution B based on current iteration times _best Updating global optimal solution G _best And its corresponding fitness function value

Meanwhile, based on worst solution B under current iteration times _bad Updating global worst solution G _bad And its corresponding fitness function value

wherein ,G_best I.e. the optimal individual in the previous T times of iteration, G _bad The worst individual in the previous T times of iteration;

s411 randomly generating a learning probability value using rand function

And f is set _rand And->

Comparing the sizes;

s412, according to f _rand And (3) with

Selecting a corresponding learning rate calculation mode to perform ignition rule probability matrix P _R Is updated by learning;

s413, increasing the value of the iteration number T by 1, and returning to the step S45;

s414, the global individual optimal solution G corresponding to the current fitness function value _best As the optimal solution when the random self-adjusting pulse nerve P system is optimized and solved;

in the step S412, when f _rand Less than

When igniting the regular probability matrix P _R The method for learning and updating comprises the following steps:

r1, randomly selecting two individuals k different from the current individual i from H solutions generated by H ESNPSs ₁ and k₂ Determining individual k ₁ and k₂ Whether the corresponding fitness function value satisfies

If yes, entering a step R2;

if not, entering a step R3;

r2, let current individual i go to individual k ₁ Learning, i.e.

Entering a step R4;

r3, let current individual i go to individual k ₂ Learning, i.e.

Entering a step R4;

wherein ,

and />

Is chromosome, is->

Is an intermediate variable +.>

and />

Respectively the kth ₁ 、k ₂ The j-th binary code of the stripe chromosome;

r4, judging the current

Whether or not to establish;

if yes, the current probability ignition rule excites the probability value

The value of increase is +.>

Realizing ignition rule probability matrix P _R Is updated by learning;

if not, the current probability ignition rule excites the probability value

The value of (2) is reduced to->

Realizing ignition rule probability matrix P _R Is updated by learning;

wherein the ignition rule excites probability values

An increase delta of (a) ₁ The calculation formula of (2) is as follows:

ignition rule excitation probability value

Delta of the decrease in (a) ₀ The calculation formula of (2) is as follows:

when f _rand Greater than

When igniting the regular probability matrix P _R The learning updating method comprises the following steps:

judging G under the current iteration number _bestj Whether or not=1 is true;

if yes, the current ignition rule excites the probability value

Increase to->

Realizing ignition rule probability matrix P _R Is updated by learning;

if not, the current ignition rule excites the probability value

Reduced to->

Realizing ignition rule probability matrix P _R Is updated by learning;

wherein the ignition rule excites probability values

An increment of delta' ₁ The calculation formula of (2) is as follows:

ignition rule excitation probability value

Delta 'of decrease in (2)' ₀ The calculation formula of (2) is as follows:

in the formula,G_bestj The j-th bit binary code of the global optimal solution searched under the current iteration times is obtained;

in the step S5, when the code of the optimal solution is 0, the corresponding suspected fault element is not faulty; when the code of the optimal solution is 1, the corresponding suspected fault element fails.

Images