CN111413565A - Intelligent power grid fault diagnosis method capable of identifying and measuring tampering attack - Google Patents

Abstract

The invention discloses an intelligent power grid fault diagnosis method capable of identifying and measuring tampering attack. Secondly, the invention establishes a measuring tampering attack recognition model based on the memory pulse neurolemma system, and effectively solves the problem of malfunction diagnosis caused by measuring tampering attack. Finally, the invention establishes a fault diagnosis model capable of comprehensively utilizing telemetering and teletraffic by utilizing the memory backtracking thought of a memory pulse neurolemma system, thereby not only changing the defect that the prior model cannot be established by utilizing the teletraffic based on a production rule, but also determining the fault type under the condition of diagnosing a fault element.

Description

Intelligent power grid fault diagnosis method capable of identifying and measuring tampering attack

Technical Field

The invention belongs to the technical field of intelligent power grid fault diagnosis, and particularly relates to a design of an intelligent power grid fault diagnosis method capable of identifying and measuring tampering attack.

Background

The false data injection attack is an important network attack means for a smart grid, and means that an attacker successfully injects false data into a data acquisition and monitoring System (SCADA) or a Wide Area Measurement System (WAMS) by using bad data detection loopholes based on residual errors in power transmission network state estimation, so that illegal purposes of modifying smart grid measurement values and state variables, controlling the operation state of the smart grid or obtaining economic benefits and the like are achieved. With the deep penetration of information communication technology in modern smart power grids, the exchange of the trend of a physical power grid and the information flow of an information grid is increasingly frequent, and modern smart power grids have been developed into power information physical fusion systems formed by fusing the physical power grid and the information grid. Because the information system has inevitable defects, bugs, faults and other conditions, a plurality of entrances can initiate network attacks, and the false data injection attacks can damage the safe and stable operation of the smart grid. The fault diagnosis is one of core contents of intelligent power grid safety control, and plays a key role in safe and stable operation of a power grid. Dummy data injection attack

As an important new network attack mode, false injection attack has an important influence on security control of a smart grid, and thus has received much attention. However, at present, the existing research almost only focuses on analyzing the influence of the false data injection attack on the safe and stable operation of the power system before the fault event, and the exploration of the influence of the false data injection attack on the fault diagnosis and recovery of the power system after the fault event is not involved. Although the existing research works on continuously trying to improve the identification and the defense of false data injection attacks in three defense lines of a power grid, according to practical engineering experience, absolutely strong and reliable defense capability cannot be guaranteed. Therefore, the smart grid fault diagnosis method considering the false data injection attack needs to be researched urgently, and has important practical significance for social economy and national security. The false data injection attack has various forms in the fault diagnosis system, and can be divided into two types according to the attack position: fraudulent data attacks and measurement tampering attacks, the attack location is shown in fig. 1.

As can be seen from fig. 1, the fraudulent attack occurs after the state estimation, and mainly acts on the network layer and the system layer, including: (1) injecting fraudulent data into the network layer by using system bugs to disguise the network layer, so that the accuracy of the data transmitted to the system layer cannot be guaranteed, and thus, the indirect attack of the power grid dispatching automation system is realized (see a position d in fig. 1); (2) injecting fraudulent data at the system level directly attacks the grid dispatching automation system (see position e in fig. 1). Therefore, fraudulent data attack influences authenticity and accuracy of fault information of the relay protection device by indirectly or directly attacking a power grid dispatching automation system of a power grid, and further prevents accurate and rapid diagnosis of fault equipment. The measurement tampering attack can act on a device layer and a network layer before state estimation, and malicious tampering on data collected by a measurement device such as an RTU (remote terminal unit, see position a in FIG. 1), a PMU (phasor measurement unit, see position b in FIG. 1) or network layer data (see position c in FIG. 1) is realized by breaking through a state estimator of a smart grid by using a bug of the smart grid. In the attack mode, the tampered data can directly act on the RTU, the PMU or the relay protection device, so that the acquired data is wrong, a fault system cannot make an effective fault decision timely and accurately, and the subsequent fault diagnosis and recovery are affected catastrophically, thereby causing great economic loss and adverse social influence. Therefore, the method for diagnosing the fault under the condition of measuring the tampering attack behavior has important significance.

Until now, a variety of excellent diagnostic methods have emerged in the field of grid fault diagnosis, such as expert systems, artificial neural networks, Bayesian networks, petri networks, optimization algorithms, and pulse neuromembrane systems. These methods all perform well when the grid is not involved in a measurement tamper attack. However, when the fault event involves a metrology tampering attack, the diagnostic correctness of the above method will be greatly reduced (often with severe misdiagnosis). Therefore, a new fault diagnosis method capable of coping with the measurement tampering attack is urgently sought.

Among the above methods, the pulse Neural membrane System (SNPS) has attracted much attention due to its powerful information processing and computing capabilities, knowledge expression and reasoning capabilities, and has been recently used to search for a brand new fault information processing mechanism and fault diagnosis method based on a membrane computing framework, and has become a research hotspot in the field of current membrane computing and biological computing. The SNPS is inspired by a mechanism that biological neurons/human brains store, transmit and exchange information in a pulse (Spike) form, and provides a special neural membrane system which is a brand-new bionic distributed intelligent parallel computing model. However, up to now, the existing SNPS fault diagnosis model is built based on a generative rule by using discrete remote semaphore. The modeling concepts described above do not take advantage of telemetry information that is continuous in the time dimension and cannot determine the type of failure of a failed component. In addition, when the measuring tampering attack is taken into consideration, the relay protection device is mistakenly operated or refused due to the tampered telemetering amount, namely, the accuracy of the telemetering amount cannot be guaranteed, and the error rate is in direct proportion to the degree of the attack. Therefore, the traditional modeling method only using the telemetric information cannot adapt to the practical problems facing the field of grid fault diagnosis in the network attack background.

In summary, the current fault diagnosis method for the smart grid has the following problems:

(1) the fault diagnosis method does not consider the influence caused by measuring the tampering attack, and the existing fault diagnosis method can cause serious misdiagnosis when the tampering attack behavior is measured.

(2) Existing fault diagnosis models based on production rules, such as: the petri net model and the SNPS model cannot comprehensively utilize telemeasurement and telesignaling to realize fault diagnosis, so that decision-making risks exist, the telesignaling subjected to tampering can cause misoperation or refusal of a relay protection device, namely the accuracy of the telesignaling cannot be guaranteed, and the fault type of a fault element cannot be judged.

Disclosure of Invention

The invention aims to solve the problems of the existing intelligent power grid fault diagnosis method and provides an intelligent power grid fault diagnosis method capable of identifying and measuring tampering attack.

The technical scheme of the invention is as follows: a smart power grid fault diagnosis method capable of identifying measurement tampering attack comprises the following steps:

and S1, aiming at the elements which have faults in the smart grid and have completed the breaker on-off detection, determining suspected fault elements by adopting an electric quantity clustering detection method.

S2, respectively establishing a measurement tampering attack recognition model based on a memory pulse neurolemma system for each suspected fault element, solving through a TRMA algorithm, judging whether each suspected fault element suffers measurement tampering attack, if so, entering step S3, and otherwise, entering step S4.

And S3, judging that the action of the relay protection device at the moment is a false action caused by measuring the tampering attack telemeasurement, and finishing the fault diagnosis of the smart power grid.

S4, aiming at each suspected fault element which is not subjected to the measurement tampering attack, respectively establishing a fault diagnosis model based on a memory pulse neural membrane system according to the topological structure and the protection pattern of the power grid, solving through an FDMA algorithm to obtain a fault diagnosis result of the suspected fault element, and finishing the fault diagnosis of the intelligent power grid.

Further, step S1 includes the following substeps:

and S11, acquiring voltage values of each bus, each generator and each transformer in 0.1S, 0.2S and 0.5S in the intelligent power grid when a relay in the intelligent power grid acts or a breaker trips.

S12, performing cluster analysis on the normalized voltage value and the normalized voltage value of the smart grid in the normal operation state, wherein the normalization formula is as follows:

wherein x' represents the normalized voltage value, x represents the voltage value to be normalized, and x_maxAnd x_minRespectively the maximum value and the minimum value in the voltage values to be normalized.

And S13, taking the element farthest from the cluster center as a suspected fault element.

Further, step S2 includes the following substeps:

s21, respectively establishing a tamper attack detection model pi based on a memory pulse neurolemma system for each suspected fault element_T。

And S22, carrying out normalization processing on the SCADA telemetering measurement and the RTU telemetering measurement of the suspected fault element at the moment of the fault to obtain a real-time fault voltage normalization value of the SCADA voltage and the RTU voltage within a sampling time range.

S23, respectively splitting the real-time fault voltage normalized values of the SCADA voltage and the RTU voltage in the sampling time range into S sampling time points, and inputting the SCADA real-time fault voltage normalized values of the sampling time points into a measuring and tampering attack identification model pi_TIn the perception neuron, the RTU real-time fault voltage normalization value of each sampling time point is input into a measurement tampering attack recognition model pi_TIn the memory neurons of (1).

S24, identifying pi through TRMA algorithm for measuring and tampering attack of each suspected fault element_TSolving to obtain various measurement tampering attack identification models pi_TAnd outputs the pulse value of the neuron.

S25, sequentially judging pi of each measured tampering attack recognition model_TIf the pulse value of any output neuron is lower than the detection threshold value, the suspected faulty element is determined to be attacked by measurement tampering, and the step S3 is entered, otherwise, the suspected faulty element is determined not to be attacked by measurement tampering, and the step S4 is entered.

Further, the measured tampering attack recognition model Π of each suspected fault element through the TRMA algorithm in step S24_TThe specific method for solving is as follows:

a1, setting the inference step number g to be 0.

A2, performing ignition calculation for each storage neuron meeting the ignition condition, and updating according to the following formula_g+1And v_g+1：

A3 for each one satisfying ignition conditionsCalculating the firing calculation of the neuron and updating theta according to the following formula_g+1And τ_g+1：

A4, adding 1 to the inference step number g.

A5, judging whether the operation condition theta is met_g≠0₁Or_g≠0₂If yes, returning to the step A2, otherwise, ending the TRMA algorithm, and outputting the measured tampering attack recognition model II_TAnd outputs the pulse value of the neuron.

The meaning of the vector, matrix and operator involved in the TRMA algorithm is as follows:

θ＝(θ₁,θ₂,...,θ_p)^Trepresenting a vector of pulse values of a storage neuron, where θ_iThe pulse value of the ith storage neuron is [0,1 ]]The real number above, i 1,2, p, p denotes the number of storage neurons; tamper-evident model pi in measurement_TIn the case where 1. ltoreq. i. ltoreq.s, θ_iFor the SCADA real-time fault voltage normalization value, when (s +1) is less than or equal to i and less than or equal to (s + s c), theta_iAnd the normalized value is the RTU real-time fault voltage, c is the total number of the memory events, and s is the total number of the sampling time points. ═ e (₁,₂,...,_q)^TRepresenting a vector of calculated neuron pulse values, wherein_jCalculating the impulse value of the neuron for the jth, and taking the value as [0,1 ]]The real number above, j 1,2,.., q, represents the number of computational neurons. τ ═ t (τ)₁,τ₂,...,τ_p)^TRepresenting a vector of storage neuron tag values, where τ_iThe tag value for the ith storage neuron is set to [0, c]The above integer; tamper-evident model pi in measurement_TIn (1) i.ltoreq (s + s c), τ is_iAll are 0. V ═ v (v)₁,ν₂,...,ν_q)^TRepresenting a computed neuron tag valueVector, where v_jCalculating the label value of the neuron for the jth, and taking the value as [0, c]The above integer. D₁＝(d_ij)_p×qIs a p × q-order matrix representing the directional synaptic connection relationship from the storage neuron to the dis-computing neuron, if the storage neuron is sigma_iTo dis computational neuron sigma_jPresence of directional synaptic connection, then d _ij1, otherwise d_ij＝0。D₂＝(d_ij)_p×qIs a p × q-order matrix representing the directional synaptic connection relationship from the storage neuron to the max computation neuron, if the storage neuron is sigma_iCalculating neuron sigma by max_jPresence of directional synaptic connection, then d _ij1, otherwise d_ij＝0。D₃＝(d_ij)_p×qIs a matrix of p × q order, which represents the directional synaptic connection relationship from the storage neuron to the min computation neuron, if the storage neuron is sigma_iCalculating neuron sigma by min_jPresence of directional synaptic connection, then d _ij1, otherwise d_ij＝0。D₄＝(d_ij)_p×qIs a matrix of p × q order, which represents the directional synaptic connection relationship from the storage neuron to the rel computation neuron, if the storage neuron is sigma_iTo rel computing neuron sigma_jPresence of directional synaptic connection, then d _ij1, otherwise d_ij＝0。E＝(e_ji)_q×pIs a q × order matrix representing the directional synaptic connection relationship from the computing neuron to the storage neuron, if the computing neuron is sigma_jTo the storage neuron σ_iExistence of synaptic connection, then e _ji1, otherwise e_ji0.Δ represents dis calculation, and D^TΔθ＝(d₁,d₂,...,d_q) Wherein d is_j＝|d_1j×θ₁-d_2j×θ₂…-d_pj×θ_pL. Denotes max calculation, and D^T·θ＝(d₁,d₂,...,d_q) Wherein d is_j＝max(d_1j×θ₁,d_2j×θ₂,...,d_pj×θ_p)。

Represents min calculation, and

wherein d is_j＝min(d_1j×θ₁,d_2j×θ₂...,d_pj×θ_p)。

Represents rel calculation, and

wherein

θ_maxRepresents the pulse value of the neuron, θ, located before the rel-computing neuron and after the max-computing neuron_disRepresenting the pulse value of the neuron before the rel computing neuron and after the dis computing neuron, wherein rho is a memory resolution coefficient, and the model pi is identified in the measurement and tampering attack recognition_TThe median value is 1.

Represents a tag-fetching calculation, an

Wherein

For tag-taking operations, if and only if

All internal non-0 elements being the same, d_j＝d_1j×τ₁Otherwise d_j＝0；

Wherein

If and only if

All internal non-0 elements being identical, e_i＝e_1i×ν₁Else, e_i＝0。

Represents a summation calculation, an

Wherein

The superscript T denotes the transpose of the vector sum matrix and the subscript g denotes the number of inference steps.

Further, step S4 includes the following substeps:

s41, respectively establishing a fault diagnosis model pi based on a memory pulse neurolemma system according to the topological structure and the protection pattern of the power grid aiming at each suspected fault element which is not subjected to the measurement tampering attack_F(ii) a Fault diagnosis model pi_FII comprises a positive sequence voltage fault diagnosis model_FPNegative sequence voltage fault diagnosis model II_FNZero sequence voltage fault diagnosis model II_FZII model for diagnosing sum remote signaling quantity fault_FB。

S42, respectively carrying out normalization processing on the SCADA telemetering value of the suspected fault element and the historical telemetering values of different types of faults of the element to obtain a SCADA real-time fault voltage normalization value and a SCADA historical fault voltage normalization value.

S43, dividing the SCADA real-time fault voltage normalization value and the SCADA historical fault voltage normalization value into S sampling time points respectively, and inputting the SCADA real-time fault voltage normalization value of each sampling time point into the positive sequence voltage fault diagnosis model pi respectively_FPNegative sequence voltage fault diagnosis model II_FNII model for diagnosing fault of sum zero sequence voltage_FZIn the sensory neuron, a remote semaphore pulse value related to the element is input to a remote semaphoreVolume fault diagnosis model pi_FBIn the perception neuron, the normalized values of the SCADA historical fault voltage at each sampling time point are respectively input into a positive sequence voltage fault diagnosis model II_FPNegative sequence voltage fault diagnosis model II_FNII model for diagnosing fault of sum zero sequence voltage_FZIn the memory neuron, a history fault remote communication quantity related to the element is inputted as a pulse value to a remote communication quantity fault diagnosis model II_FBIn the memory neurons of (1).

S44, carrying out fault diagnosis model pi on each suspected fault element through FDMA algorithm_FSolving to obtain various fault diagnosis models pi_FThe pulse value and the memory tag value of the neuron are output.

S45, calculating each fault diagnosis model pi_FThe average value f of the pulse values corresponding to the memory tag value_zAnd f is_zThe basic fault event represented by the memory label value corresponding to the maximum value is used as the fault diagnosis result of the suspected fault element, and the pulse value mean value f corresponding to the memory label value_zThe calculation formula of (2) is as follows:

wherein z represents the memory tag value of the output neuron, and the value is [0, c ]]The whole number of (a) to (b),

II representing positive sequence voltage fault diagnosis model_FPThe corresponding pulse value when the memory label value of the middle output neuron is z,

negative sequence voltage fault diagnosis model II_FNThe corresponding pulse value when the memory label value of the middle output neuron is z,

zero sequence voltage fault diagnosis model II_FZThe corresponding pulse value when the memory label value of the middle output neuron is z,

II model for representing remote signalling quantity fault diagnosis_FBThe corresponding pulse value when the memory tag value of the middle output neuron is 1,

II model for representing remote signalling quantity fault diagnosis_FBAnd when the memory tag value of the middle output neuron is 2, the corresponding pulse value is obtained.

Further, the fault diagnosis model pi for each suspected fault element through FDMA algorithm in step S44_FThe specific method for solving is as follows:

b1, setting the inference step number g as 0.

B2, performing ignition calculation on each storage neuron meeting the ignition condition, and updating according to the following formula_g+1And v_g+1：

B3, performing ignition calculation on each calculation neuron meeting the ignition condition, and updating theta according to the following formula_g+1And τ_g+1：

B4, adding 1 to the inference step number g.

B5, judging whether the operation condition theta is met_g≠0₁Or_g≠0₂If yes, returning to the step A2, otherwise, ending the FDMA algorithm, and outputting to obtain each fault diagnosis model II_FPulse value sum of intermediate output neuronsThe tag value is recalled.

The vector, matrix and operator involved in the FDMA algorithm have the following meanings:

θ＝(θ₁,θ₂,...,θ_p)^Trepresenting a vector of pulse values of a storage neuron, where θ_iThe pulse value of the ith storage neuron is [0,1 ]]The real number above, i 1,2, p, p denotes the number of storage neurons; II model for fault diagnosis_FIn the case where 1. ltoreq. i. ltoreq.s, θ_iFor the SCADA real-time fault voltage normalization value, when (s +1) is less than or equal to i and less than or equal to (s + s c), theta_iThe method comprises the steps of obtaining SCADA historical fault voltage normalization values of different fault types, obtaining c the total number of memory events, and obtaining s the total number of sampling time points. ═ e (₁,₂,...,_q)^TRepresenting a vector of calculated neuron pulse values, wherein_jCalculating the impulse value of the neuron for the jth, and taking the value as [0,1 ]]The real number above, j 1,2,.., q, represents the number of computational neurons. τ ═ t (τ)₁,τ₂,...,τ_p)^TRepresenting a vector of storage neuron tag values, where τ_iThe tag value for the ith storage neuron is set to [0, c]The above integer; II model for fault diagnosis_FIn the case where 1. ltoreq. i.ltoreq.s, τ_iAll are 0, and when (s +1) ≦ i ≦ (s + s × c), τ is_iTo remember the basic failure event type. V ═ v (v)₁,ν₂,...,ν_q)^TRepresenting a vector of computed neuron tag values, where v_jCalculating the label value of the neuron for the jth, and taking the value as [0, c]The above integer. D₁＝(d_ij)_p×qIs a p × q-order matrix representing the directional synaptic connection relationship from the storage neuron to the dis-computing neuron, if the storage neuron is sigma_iTo dis computational neuron sigma_jPresence of directional synaptic connection, then d_ij1, otherwise d_ij＝0。D₂＝(d_ij)_p×qIs a p × q-order matrix representing the directional synaptic connection relationship from the storage neuron to the max computation neuron, if the storage neuron is sigma_iCalculating neuron sigma by max_jPresence of directional synaptic connection, then d_ij1, otherwise d_ij＝0。D₃＝(d_ij)_p×qIs a matrix of p × q order, which represents the directional synaptic connection relationship from the storage neuron to the min computation neuron, if the storage neuron is sigma_iCalculating neuron sigma by min_jPresence of directional synaptic connection, then d_ij1, otherwise d_ij＝0。D₄＝(d_ij)_p×qIs a matrix of p × q order, which represents the directional synaptic connection relationship from the storage neuron to the rel computation neuron, if the storage neuron is sigma_iTo rel computing neuron sigma_jPresence of directional synaptic connection, then d_ij1, otherwise d_ij＝0。E＝(e_ji)_q×pIs a q × order matrix representing the directional synaptic connection relationship from the computing neuron to the storage neuron, if the computing neuron is sigma_jTo the storage neuron σ_iExistence of synaptic connection, then e_ji1, otherwise e_ji0.Δ represents dis calculation, and D^TΔθ＝(d₁,d₂,...,d_q) Wherein d is_j＝|d_1j×θ₁-d_2j×θ₂…-d_pj×θ_pL. Denotes max calculation, and D^T·θ＝(d₁,d₂,...,d_q) Wherein d is_j＝max(d_1j×θ₁,d_2j×θ₂,...,d_pj×θ_p)。

Represents min calculation, and

wherein d is_j＝min(d_1j×θ₁,d_2j×θ₂...,d_pj×θ_p)。

Represents rel calculation, and

wherein

θ_maxRepresents the pulse value of the neuron, θ, located before the rel-computing neuron and after the max-computing neuron_disRepresenting the pulse value of the neuron before the rel calculating neuron and after the dis calculating neuron, wherein rho is a memory resolution coefficient and is in a fault diagnosis pi model_FThe median value is 0.5.

Represents a tag-fetching calculation, an

Wherein

For tag-taking operations, if and only if

All internal non-0 elements being the same, d_j＝d_1j×τ₁Otherwise d_j＝0；

Wherein

If and only if

All internal non-0 elements being identical, e_i＝e_1i×ν₁Else, e_i＝0。

Represents a summation calculation, an

Wherein

The superscript T denotes the transpose of the vector sum matrix and the subscript g denotes the number of inference steps.

The invention has the beneficial effects that:

(1) the method determines the suspected fault element by using an electric quantity clustering detection method, thereby reducing the diagnosis range.

(2) The invention establishes a tamper attack detection model based on the memory pulse neurolemma system, and effectively solves the problem of fault diagnosis misoperation caused by tamper attack detection.

(3) The invention establishes a fault diagnosis model capable of comprehensively utilizing telemetering and teletraffic by utilizing the memory backtracking idea of a memory pulse neurolemma system, not only overcomes the defect that the prior model cannot be established by utilizing the teletraffic based on a production rule, but also can determine the fault type under the condition of diagnosing a fault element.

Drawings

Fig. 1 is a schematic view illustrating an attack position of a false data injection attack in a fault diagnosis system according to the background art of the present invention.

Fig. 2 is a flowchart of a smart grid fault diagnosis method capable of identifying a measurement tampering attack according to an embodiment of the present invention.

Fig. 3 is a schematic diagram of a sensing neuron and a simplified representation thereof according to an embodiment of the present invention.

FIG. 4 is a schematic diagram of a memory neuron and a simplified representation thereof according to an embodiment of the present invention.

Fig. 5 is a schematic diagram of a transmission neuron and a simplified representation thereof according to an embodiment of the present invention.

Fig. 6 is a schematic diagram illustrating a dis computing neuron and a simplified representation thereof according to an embodiment of the present invention.

Fig. 7 is a schematic diagram of a max computing neuron and a simplified representation thereof according to an embodiment of the present invention.

Fig. 8 is a schematic diagram of a min computation neuron and a simplified representation thereof according to an embodiment of the present invention.

FIG. 9 is a schematic diagram of a rel computing neuron and its simplified form according to an embodiment of the present invention.

Fig. 10 is a schematic diagram of a measured tampering attack recognition model based on a memory pulse neurolemma system according to an embodiment of the present invention.

Fig. 11 is a schematic diagram of a fault diagnosis model based on a memory impulse neural membrane system according to an embodiment of the present invention.

Fig. 12 is a schematic diagram illustrating a clustering result of suspected fault areas according to an embodiment of the present invention.

Fig. 13 is a schematic diagram of a pulse value curve of a sensing neuron and a memory neuron in the tamper attack recognition model measured by the line L1213 according to an embodiment of the present invention.

Fig. 14 is a schematic diagram illustrating pulse values of output neurons in the line L1213 measured tampering attack recognition model according to an embodiment of the present invention.

Fig. 15 is a schematic diagram illustrating a detection result of a tamper attack in device measurement according to an embodiment of the present invention.

Detailed Description

Exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It is to be understood that the embodiments shown and described in the drawings are merely exemplary and are intended to illustrate the principles and spirit of the invention, not to limit the scope of the invention.

The embodiment of the invention provides an intelligent power grid fault diagnosis method capable of identifying and measuring tampering attack, which comprises the following steps of S1-S4 as shown in FIG. 2:

and S1, aiming at the elements which have faults in the smart grid and have completed the breaker on-off detection, determining suspected fault elements by adopting an electric quantity clustering detection method.

The step S1 includes the following substeps S11-S13:

and S11, acquiring voltage values of each bus, each generator and each transformer in 0.1S, 0.2S and 0.5S in the intelligent power grid when a relay in the intelligent power grid acts or a breaker trips.

S12, performing cluster analysis on the normalized voltage value and the normalized voltage value of the smart grid in the normal operation state, wherein the normalization formula is as follows:

wherein x' represents the normalized voltage value, x represents the voltage value to be normalized, and x_maxAnd x_minRespectively the maximum value and the minimum value in the voltage values to be normalized.

And S13, taking the element farthest from the cluster center as a suspected fault element.

S2, respectively establishing a measurement tampering attack recognition model based on a memory pulse neurolemma system for each suspected fault element, solving through a TRMA algorithm, judging whether each suspected fault element suffers measurement tampering attack, if so, entering step S3, and otherwise, entering step S4.

The step S2 includes the following substeps S21-S25:

s21, respectively establishing a tamper attack detection model pi based on a memory pulse neurolemma system for each suspected fault element_T。

In the embodiment of the invention, a measuring, tampering, attacking and identifying model II based on a memory pulse neurolemma system_TThe method specifically comprises the following steps:

Π_T＝(O,σ₁,...,σ_m,syn,in,out)

where O ═ { a } represents a set of nerve pulses, a represents a nerve pulse, σ represents a set of nerve pulses, and₁,...,σ_midentifying model Π for measuring tampering attacks_TM neurons of (1), σ_i＝(θ_i,λ_i,τ_i,r_i),i＝1,2,...,m，θ_iThe pulse value of the ith neuron is represented by [0,1 ]]The real number of (c); lambda [ alpha ]_iExpressing the ignition threshold value of the ith neuron, wherein the value of the ignition threshold value is any real number; tau is_iThe memory tag value of the ith neuron is [0, c ]]C is the total number of memory events; specifically, when no tag value exists in a neuron, the tag value is 0; r is_iIndicating the firing rule of the ith neuron, in the form of E/a^(θ,τ)→a^(β,τ)Wherein E ═ aⁿ,θ≥λ_iIs the ignition conditionDenotes the condition if and only if the neuron σ_iReceiving at least n pulses and the neuron pulse value satisfies theta ≧ lambda_iThe firing rule can only be executed, at which time the neuron σ_iA pulse with potential value theta is consumed, a new pulse with potential value β is generated and transmitted backwards, and at the same time, the label value tau of the neuron_iIs consumed and passes a new tag value tau backward_iOtherwise, the neuron does not execute ignition calculation, syn {1, 2.. multidot.m } × {1, 2.. multidot.m } represents directed synapse connection relation between neurons, and when i is more than or equal to 1 and less than or equal to m, the condition is satisfied

in and out respectively represent a measured tampering attack recognition model pi_TA set of input neurons and a set of output neurons.

In the embodiment of the invention, a tampering attack identification model II is measured_TTamper attack identification model pi including positive sequence voltage measurement_TPNegative sequence voltage measurement tampering attack identification model II_TNTampering attack recognition model pi with zero sequence voltage measurement_TZ。

In the embodiment of the invention, a tampering attack identification model II is measured_TIncludes a storage neuron for sensing or storing the pulse value and the memory event tag value from the environment and a computation neuron for computing the pulse value and the memory event tag value.

The storage neurons include sensing neurons, memory neurons and transmission neurons, as shown in fig. 3 to 5, which are respectively represented by circles, triangles and diamonds, pulse values generated after all types of storage neurons are ignited satisfy β ═ θ, and specific functions are respectively described as follows:

(1) the sensing neurons are used to acquire the detection information (such as teletraffic and telemeasurement) from the environment, convert it into real-time basic event features, and transmit them backward in the form of neural impulse values, as shown in fig. 3. The perception neuron is unable to store the memory event signature, i.e., it does not contain an event tag. Therefore, the pulse value θ represents the real-time basic event characteristic value of the current event, and the memory tag value is 0.II for measuring tampering attack recognition model provided by the embodiment of the invention_TThe pulse values theta of the perception neurons represent fault voltage normalization values of corresponding sampling time points in the power grid SCADA system; the memory tags of the sensory neurons are all 0.

(2) The memory neurons are used for storing the memory basic event features and the corresponding memory tag values, as shown in fig. 4. Tamper-evident model pi in measurement_TEach memory neuron corresponds to a real-time basic fault event characteristic, so that the pulse value theta of each memory neuron corresponds to a characteristic value when the real-time basic fault event occurs, namely a fault voltage normalization value corresponding to a sampling time point in an RTU under the current fault condition; the memory tag cannot be determined, and is therefore collectively denoted as 0.

(3) The transmission neuron is used to transmit the impulse value and the memory tag value transmitted from its pre-synaptic neuron to its post-synaptic neuron, as shown in fig. 5. Specifically, if the transmission neuron is not present as a post-synaptic neuron, it is an output neuron responsible for outputting the pulse value and the memory tag value into the environment.

The quantity relation of the various storage neurons meets the following requirements: if the number of storage neurons in a memory pulse neurolemma system is p, wherein the number of sensing neurons is s, and the number of memory events is c, the number of memory neurons is s × c, and the number of output neurons is p-s-s × c.

The calculating neurons comprise dis calculating neurons, max calculating neurons, min calculating neurons and rel calculating neurons, 4 types of calculating neurons have the same functions, namely pulse values and memory tag values are transmitted to the postsynaptic transmission neurons according to the ignition rules, and the functions of different calculating neurons are described as follows:

(1) as shown in FIG. 6, the tampering attack recognition model pi is measured_TAnd the dis calculating neuron is used for acquiring the matching degree between the current SCADA real-time basic fault feature and the RTU real-time basic fault feature by calculating the difference degree of pulse values of the sensing neuron and the memory neuron.

(2) As shown in fig. 7, the max calculating neuron is used to logically or the impulse value transmitted by its pre-synaptic neuron, i.e. the maximum impulse value of its pre-synaptic neuron is taken as the impulse value of the current max calculating neuron to participate in the subsequent calculation.

(3) As shown in fig. 8, the min computation neuron is configured to perform a logical and computation on the impulse values transmitted by its pre-synaptic neurons, that is, to take the minimum impulse value of its pre-synaptic neuron as the impulse value of the current min computation neuron to participate in the subsequent computation.

(4) As shown in fig. 9, the rel computation neuron is configured to perform global matching calculation on the impulse values transmitted by its pre-synaptic neurons, i.e. find the global matching degree of the transmission neuron with the smallest number in its pre-synaptic neurons. Theta in FIG. 9_minRepresents the pulse value of the neuron before rel and after min_maxRepresents the pulse value of the neuron, θ, located before the rel-computing neuron and after the max-computing neuron_disRepresenting the pulse value of the neuron before the rel computing neuron and after the dis computing neuron, wherein rho is a memory resolution coefficient, and the model pi is identified in the measurement and tampering attack recognition_TThe median value is 1.

In FIGS. 6 to 9

In order to obtain the tag value, the specific operation mode is as follows:

where k represents the total number of neurons performing the tag-fetching operation, τ_iThe memory tag value of the ith neuron that performs the tag operation.

Respectively establishing a measuring tampering attack recognition model pi by using the neurons_TStoring RTU telemetering data into Memory neuron, sending SCADA telemetering data into sensing neuron, outputting output neuron pulse value of Memory pulse neural membrane System (MSNPS) by Memory backtracking mechanism, judging whether initial RTU telemetering value is consistent with SCADA telemetering value, and finally completing measurementTamper attack identification. Establishing a II model for measuring, tampering, attacking and identifying based on a memory pulse neurolemma system according to the definition of various neurons_T(MSNPS-TRM) as shown in FIG. 10.

II model for measuring and identifying tampering_T＝(O,σ₁,...,σ_mSyn, in, out), the sensing neuron pulse value is a fault voltage normalization value within a sampling time range in the SCADA. The method specifically comprises the following steps: dividing a fault voltage normalization value of a suspected fault element in a target power grid SCADA within a certain sampling time range into s sampling time points, and sending the fault voltage normalization value of each sampling time point to s corresponding sensing neurons as pulse values of the sampling time points; at this time, the perception neuron does not contain the value of the memory tag, i.e., τ_i0 (1. ltoreq. i. ltoreq. s). The pulse value of the memory neuron is a failure voltage normalization value in a sampling time range in an RTU, and the acquisition mode of the pulse value of the memory neuron is the same as that of the pulse value of the sensing neuron, but the source of the pulse value is different. At this time, measuring the tampering identification model II_TThe number of memory events stored by the internal memory neuron is 0(c ═ 0), and the corresponding memory tag value is 0, i.e., τ_i0(s + 1. ltoreq. i.ltoreq.2s). Tamper attack identification model pi for positive sequence voltage measurement_TPNegative sequence voltage measurement tampering attack identification model II_TNTampering attack recognition model pi with zero sequence voltage measurement_TZAnd respectively selecting the positive sequence, negative sequence and zero sequence voltage normalization values as pulse values of the sensing neuron and the memory neuron.

And S22, carrying out normalization processing on the SCADA telemetering measurement and the RTU telemetering measurement of the suspected fault element at the moment of the fault to obtain a real-time fault voltage normalization value of the SCADA voltage and the RTU voltage within a sampling time range.

S23, respectively splitting the real-time fault voltage normalized values of the SCADA voltage and the RTU voltage in the sampling time range into S sampling time points, and inputting the SCADA real-time fault voltage normalized values of the sampling time points into a measuring and tampering attack identification model pi_TIn the perception neuron, the RTU real-time fault voltage normalization value of each sampling time point is input into a measurement tampering attack recognition model pi_TIn the memory neurons of (1).

S24, identifying pi through TRMA algorithm for measuring and tampering attack of each suspected fault element_TSolving to obtain various measurement tampering attack identification models pi_TAnd outputs the pulse value of the neuron.

Measuring tampering attack recognition model pi of each suspected fault element through TRMA algorithm_TThe specific method for solving is as follows:

a1, setting the inference step number g to be 0.

A2, performing ignition calculation for each storage neuron meeting the ignition condition, and updating according to the following formula_g+1And v_g+1：

A3, performing ignition calculation for each calculation neuron meeting the ignition condition, and updating theta according to the following formula_g+1And τ_g+1：

A4, adding 1 to the inference step number g.

A5, judging whether the operation condition theta is met_g≠0₁Or_g≠0₂If yes, returning to the step A2, otherwise, ending the TRMA algorithm, and outputting the measured tampering attack recognition model II_TAnd outputs the pulse value of the neuron.

The meaning of the vector, matrix and operator involved in the TRMA algorithm is as follows:

θ＝(θ₁,θ₂,...,θ_p)^Trepresenting a vector of pulse values of a storage neuron, where θ_iIs the ithStoring the pulse value of the neuron as [0,1 ]]The real number above, i 1,2, p, p denotes the number of storage neurons; tamper-evident model pi in measurement_TIn the case where 1. ltoreq. i. ltoreq.s, θ_iFor the SCADA real-time fault voltage normalization value, when (s +1) is less than or equal to i and less than or equal to (s + s c), theta_iAnd the normalized value is the RTU real-time fault voltage, c is the total number of the memory events, and s is the total number of the sampling time points.

＝(₁,₂,...,_q)^TRepresenting a vector of calculated neuron pulse values, wherein_jCalculating the impulse value of the neuron for the jth, and taking the value as [0,1 ]]The real number above, j 1,2,.., q, represents the number of computational neurons.

τ＝(τ₁,τ₂,...,τ_p)^TRepresenting a vector of storage neuron tag values, where τ_iThe tag value for the ith storage neuron is set to [0, c]The above integer; tamper-evident model pi in measurement_TIn (1) i.ltoreq (s + s c), τ is_iAll are 0.

ν＝(ν₁,ν₂,...,ν_q)^TRepresenting a vector of computed neuron tag values, where v_jCalculating the label value of the neuron for the jth, and taking the value as [0, c]The above integer.

D₁＝(d_ij)_p×qIs a p × q-order matrix representing the directional synaptic connection relationship from the storage neuron to the dis-computing neuron, if the storage neuron is sigma_iTo dis computational neuron sigma_jPresence of directional synaptic connection, then d _ij1, otherwise d_ij＝0。

D₂＝(d_ij)_p×qIs a p × q-order matrix representing the directional synaptic connection relationship from the storage neuron to the max computation neuron, if the storage neuron is sigma_iCalculating neuron sigma by max_jPresence of directional synaptic connection, then d _ij1, otherwise d_ij＝0。

D₃＝(d_ij)_p×qIs a matrix of order p × q, which represents the directed synaptic connection relationship from the storage neuron to the min computation neuron, if anyStore neuron sigma_iCalculating neuron sigma by min_jPresence of directional synaptic connection, then d _ij1, otherwise d_ij＝0。

D₄＝(d_ij)_p×qIs a matrix of p × q order, which represents the directional synaptic connection relationship from the storage neuron to the rel computation neuron, if the storage neuron is sigma_iTo rel computing neuron sigma_jPresence of directional synaptic connection, then d _ij1, otherwise d_ij＝0。

E＝(e_ji)_q×pIs a q × order matrix representing the directional synaptic connection relationship from the computing neuron to the storage neuron, if the computing neuron is sigma_jTo the storage neuron σ_iExistence of synaptic connection, then e _ji1, otherwise e_ji＝0。

Δ represents dis calculation, and D^TΔθ＝(d₁,d₂,...,d_q) Wherein d is_j＝|d_1j×θ₁-d_2j×θ₂...-d_pj×θ_p|。

Denotes max calculation, and D^T·θ＝(d₁,d₂,...,d_q) Wherein d is_j＝max(d_1j×θ₁,d_2j×θ₂,...,d_pj×θ_p)。

Represents min calculation, and

wherein d is_j＝min(d_1j×θ₁,d_2j×θ₂...,d_pj×θ_p)。

Represents rel calculation, and

wherein

θ_maxRepresents the pulse value of the neuron, θ, located before the rel-computing neuron and after the max-computing neuron_disRepresenting the pulse value of the neuron before the rel computing neuron and after the dis computing neuron, wherein rho is a memory resolution coefficient, and the model pi is identified in the measurement and tampering attack recognition_TThe median value is 1. Specially, equivalent measurement tampering attack recognition model II_TWhen there is only one memory event tag, i.e. there are only s sensing neurons and s memory neurons,

represents a tag-fetching calculation, an

Wherein

For tag-taking operations, if and only if

All internal non-0 elements being the same, d_j＝d_1j×τ₁Otherwise d_j＝0；

Wherein

If and only if

All internal non-0 elements being identical, e_i＝e_1i×ν₁Else, e_i＝0。

Represents a summation calculation, an

Wherein

The superscript T denotes the transpose of the vector sum matrix and the subscript g denotes the number of inference steps.

S25, sequentially judging pi of each measured tampering attack recognition model_TIf the pulse value of any output neuron is lower than the detection threshold (in the embodiment of the present invention, the detection threshold is 0.8), if so, it is determined that the suspected faulty element is under the tamper attack, and the process goes to step S3, otherwise, it is determined that the suspected faulty element is not under the tamper attack, and the process goes to step S4.

And S3, judging that the action of the relay protection device at the moment is a false action caused by measuring the tampering attack telemeasurement, and finishing the fault diagnosis of the smart power grid.

S4, aiming at each suspected fault element which is not subjected to the measurement tampering attack, respectively establishing a fault diagnosis model based on a memory pulse neural membrane system according to the topological structure and the protection pattern of the power grid, solving through an FDMA algorithm to obtain a fault diagnosis result of the suspected fault element, and finishing the fault diagnosis of the intelligent power grid.

The step S4 includes the following substeps S41-S45:

s41, respectively establishing a fault diagnosis model pi based on a memory pulse neurolemma system according to the topological structure and the protection pattern of the power grid aiming at each suspected fault element which is not subjected to the measurement tampering attack_F。

Wherein, the fault diagnosis model II_FII comprises a positive sequence voltage fault diagnosis model_FPNegative sequence ofVoltage fault diagnosis model pi_FNZero sequence voltage fault diagnosis model II_FZII model for diagnosing sum remote signaling quantity fault_FB。

In the embodiment of the invention, a fault diagnosis model II based on a memory pulse neurolemma system_FThe method specifically comprises the following steps:

Π_F＝(O,σ₁,...,σ_m,syn,in,out)

where O ═ { a } represents a set of nerve pulses, a represents a nerve pulse, σ represents a set of nerve pulses, and₁,...,σ_mmodel pi for fault diagnosis_FM neurons of (1), σ_i＝(θ_i,λ_i,τ_i,r_i),i＝1,2,...,m，θ_iThe pulse value of the ith neuron is represented by [0,1 ]]The real number of (c); lambda [ alpha ]_iExpressing the ignition threshold value of the ith neuron, wherein the value of the ignition threshold value is any real number; tau is_iThe memory tag value of the ith neuron is [0, c ]]C is the total number of memory events; specifically, when no tag value exists in a neuron, the tag value is 0; r is_iIndicating the firing rule of the ith neuron, in the form of E/a^(θ,τ)→a^(β,τ)Wherein E ═ aⁿ,θ≥λ_iIs the firing condition, meaning if and only if the neuron σ_iReceiving at least n pulses and the neuron pulse value satisfies theta ≧ lambda_iThe firing rule can only be executed, at which time the neuron σ_iA pulse with potential value theta is consumed, a new pulse with potential value β is generated and transmitted backwards, and at the same time, the label value tau of the neuron_iIs consumed and passes a new tag value tau backward_iOtherwise, the neuron does not execute ignition calculation, syn {1, 2.. multidot.m } × {1, 2.. multidot.m } represents directed synapse connection relation between neurons, and when i is more than or equal to 1 and less than or equal to m, the condition is satisfied

in and out respectively represent fault diagnosis models II_FA set of input neurons and a set of output neurons.

Fault diagnosis model pi_FThe neurons in (A) includeA storage neuron for sensing or storing the pulse value and the memory event tag value from the environment and a computation neuron for computing the pulse value and the memory event tag value.

The storage neurons include sensing neurons, memory neurons and transmission neurons, as shown in fig. 3 to 5, which are respectively represented by circles, triangles and diamonds, pulse values generated after all types of storage neurons are ignited satisfy β ═ θ, and specific functions are respectively described as follows:

(1) the sensing neurons are used to acquire the detection information (such as teletraffic and telemeasurement) from the environment, convert it into real-time basic event features, and transmit them backward in the form of neural impulse values, as shown in fig. 3. The perception neuron is unable to store the memory event signature, i.e., it does not contain an event tag. Therefore, the pulse value θ represents the real-time basic event characteristic value of the current event, and the memory tag value is 0. II for fault diagnosis model provided by the embodiment of the invention_FThe pulse values theta of the perception neurons represent fault voltage normalization values of corresponding sampling time points in the power grid SCADA system; the memory tags of the sensory neurons are all 0.

(2) The memory neurons are used for storing the memory basic event features and the corresponding memory tag values, as shown in fig. 4. II model for fault diagnosis_FEach memory neuron corresponds to a memory basic failure event characteristic. Therefore, the pulse value theta corresponds to a characteristic value when the memory basic fault event occurs, namely a voltage normalization value of a sampling time point corresponding to the fault type in the SCADA system under the historical fault condition; its memory tag value corresponds to the memory base failure event type. Specifically, in the embodiment of the present invention, according to the historical fault types, it is determined that the number of the memory basic fault event types is 11 (c is 11), and the determination is: no Fault (No Fault), A-phase short-circuit Fault (A), B-phase short-circuit Fault (B), C-phase short-circuit Fault (C), AB-phase short-circuit Fault (AB), AC-phase short-circuit Fault (AC), BC-phase short-circuit Fault (BC), AB-phase ground short-circuit Fault (AB-G), AC-phase ground short-circuit Fault (AC-G), BC-phase ground short-circuit Fault (BC-G) and ABC-phase short-circuit Fault (ABC), and the memory label values thereof are sequentially represented by 1 to 11.

(3) The transmission neuron is used to transmit the impulse value and the memory tag value transmitted from its pre-synaptic neuron to its post-synaptic neuron, as shown in fig. 5. Specifically, if the transmission neuron is not present as a post-synaptic neuron, it is an output neuron responsible for outputting the pulse value and the memory tag value into the environment.

The quantity relation of the various storage neurons meets the following requirements: if the number of storage neurons in a memory pulse neurolemma system is p, wherein the number of sensing neurons is s, and the number of memory events is c, the number of memory neurons is s × c, and the number of output neurons is p-s-s × c.

The calculating neurons comprise dis calculating neurons, max calculating neurons, min calculating neurons and rel calculating neurons, 4 types of calculating neurons have the same functions, namely pulse values and memory tag values are transmitted to the postsynaptic transmission neurons according to the ignition rules, and the functions of different calculating neurons are described as follows:

(1) as shown in fig. 6, in the fault diagnosis model Π_FAnd the dis calculating neuron is used for acquiring the matching degree between the current real-time basic fault feature and the historical fault feature by calculating the difference degree of pulse values of the sensing neuron and the memory neuron.

(2) As shown in fig. 7, the max calculating neuron is used to logically or the impulse value transmitted by its pre-synaptic neuron, i.e. the maximum impulse value of its pre-synaptic neuron is taken as the impulse value of the current max calculating neuron to participate in the subsequent calculation.

(3) As shown in fig. 8, the min computation neuron is configured to perform a logical and computation on the impulse values transmitted by its pre-synaptic neurons, that is, to take the minimum impulse value of its pre-synaptic neuron as the impulse value of the current min computation neuron to participate in the subsequent computation.

(4) As shown in fig. 9, the rel computation neuron is configured to perform global matching calculation on the impulse values transmitted by its pre-synaptic neurons, i.e. find the global matching degree of the transmission neuron with the smallest number in its pre-synaptic neurons. Theta in FIG. 9_minRepresents the pulse value of the neuron before rel and after min_maxRepresenting the max-computing nerve that precedes the rel-computing neuronPulse value of neuron after neuron, theta_disRepresenting the pulse value of the neuron before the rel computing neuron and after the dis computing neuron, wherein rho is a memory resolution coefficient, and the model pi is identified in the measurement and tampering attack recognition_TThe median value is 1.

In FIGS. 6 to 9

In order to obtain the tag value, the specific operation mode is as follows:

where k represents the total number of neurons performing the tag-fetching operation, τ_iThe memory tag value of the ith neuron that performs the tag operation.

Establishing fault diagnosis model pi by using the neurons respectively_FEstablishing a fault diagnosis model pi based on a memory pulse neurolemma system according to a memory backtracking mechanism of MSNPS by utilizing the matching degree of SCADA fault telemetering data and remote signaling data under the condition of historical faults when the faults occur_F(MSNPS-FDM), the faulty element and type identification is done as shown in fig. 11.

Complete fault diagnosis model pi_F＝(O,σ₁,...,σ_mSyn, in, out) needs to be diagnosed by a positive sequence voltage fault model pi_FPNegative sequence voltage fault diagnosis model II_FNZero sequence voltage fault diagnosis model II_FZII model for diagnosing sum remote signaling quantity fault_FBAnd the four submodels are formed. Particularly, when the suspected fault element is determined to be attacked by measurement tampering, the reliability of the fault remote signaling quantity is not high, so that the remote signaling quantity fault diagnosis model II_FBAnd does not participate in subsequent calculations. It should be noted here that the remote traffic fault diagnosis model Π_FBThe impulse value of the sensory neuron represents remote signaling information corresponding to a suspected fault element (the protection device acts, the value of the impulse value is 1, and otherwise, the value of the impulse value is 0); the pulse value of the neuron is the remote information of the element under the historical fault condition (the protection device acts, the value of the protection device takes 1, otherwise takes 0), so the number of basic events is memorizedThe number is 2(c is 2), and the memory tag pulse values of no failure and failure are indicated by 1 and 2, respectively. In telemetering measuring model II_FP、Π_FNII_FZIn the method, a fault voltage normalization value of a suspected fault element in the SCADA within a certain sampling time range and a fault voltage normalization value of the element under the historical fault condition are divided into s sampling time points, and the fault voltage normalization value of each sampling time point is respectively sent to corresponding s sensing neurons and memory neurons to serve as pulse values of the sensing neurons and the memory neurons. Wherein the memory tag value of a memory neuron is related to the historical failure type of the stored telemetry measurement of that neuron.

S42, respectively carrying out normalization processing on the SCADA telemetering value of the suspected fault element and the historical telemetering values of different types of faults of the element to obtain a SCADA real-time fault voltage normalization value and a SCADA historical fault voltage normalization value.

S43, dividing the SCADA real-time fault voltage normalization value and the SCADA historical fault voltage normalization value into S sampling time points respectively, and inputting the SCADA real-time fault voltage normalization value of each sampling time point into the positive sequence voltage fault diagnosis model pi respectively_FPNegative sequence voltage fault diagnosis model II_FNII model for diagnosing fault of sum zero sequence voltage_FZIn the sensing neuron, a remote traffic pulse value related to the element is input into a remote traffic fault diagnosis model pi_FBPi (if the protection device acts, the corresponding pulse value is 1, otherwise 0) in the perception neuron_FBThe tag value of the sensory neuron is 0. Respectively inputting the SCADA historical fault voltage normalization values of all sampling time points into a positive sequence voltage fault diagnosis model II_FPNegative sequence voltage fault diagnosis model II_FNII model for diagnosing fault of sum zero sequence voltage_FZThe memory label value is the corresponding basic fault event type (1-11), and the historical fault remote communication quantity related to the element is input into a remote communication quantity fault diagnosis model pi as a pulse value_FBIf the protection device acts, the corresponding pulse value is 1, otherwise, 0 is taken, and at the moment, the memory tag value is the historical fault occurrence state corresponding to the element (if the fault occurs, the tag value is 1, otherwise, 0 is taken)

S44, carrying out fault diagnosis model pi on each suspected fault element through FDMA algorithm_FSolving to obtain various fault diagnosis models pi_FThe pulse value and the memory tag value of the neuron are output.

Fault diagnosis model pi for each suspected fault element through FDMA algorithm_FThe specific method for solving is as follows:

b1, setting the inference step number g as 0.

B2, performing ignition calculation on each storage neuron meeting the ignition condition, and updating according to the following formula_g+1And v_g+1：

B3, performing ignition calculation on each calculation neuron meeting the ignition condition, and updating theta according to the following formula_g+1And τ_g+1：

B4, adding 1 to the inference step number g.

B5, judging whether the operation condition theta is met_g≠0₁Or_g≠0₂If yes, returning to the step A2, otherwise, ending the FDMA algorithm, and outputting to obtain each fault diagnosis model II_FThe pulse value and the memory tag value of the neuron are output.

The vector, matrix and operator involved in the FDMA algorithm have the following meanings:

θ＝(θ₁,θ₂,...,θ_p)^Trepresenting a vector of pulse values of a storage neuron, where θ_iThe pulse value of the ith storage neuron is [0,1 ]]The real number above, i 1,2, p, p denotes the number of storage neurons; II model for fault diagnosis_FIn the case where 1. ltoreq. i. ltoreq.s, θ_iFor the SCADA real-time fault voltage normalization value, when (s +1) is less than or equal to i and less than or equal to (s + s c), theta_iThe method comprises the steps of obtaining SCADA historical fault voltage normalization values of different fault types, obtaining c the total number of memory events, and obtaining s the total number of sampling time points.

＝(₁,₂,...,_q)^TRepresenting a vector of calculated neuron pulse values, wherein_jCalculating the impulse value of the neuron for the jth, and taking the value as [0,1 ]]The real number above, j 1,2,.., q, represents the number of computational neurons.

τ＝(τ₁,τ₂,...,τ_p)^TRepresenting a vector of storage neuron tag values, where τ_iThe tag value for the ith storage neuron is set to [0, c]The above integer; II model for fault diagnosis_FIn the case where 1. ltoreq. i.ltoreq.s, τ_iAll are 0, and when (s +1) ≦ i ≦ (s + s × c), τ is_iTo remember the basic failure event type.

ν＝(ν₁,ν₂,...,ν_q)^TRepresenting a vector of computed neuron tag values, where v_jCalculating the label value of the neuron for the jth, and taking the value as [0, c]The above integer.

D₁＝(d_ij)_p×qIs a p × q-order matrix representing the directional synaptic connection relationship from the storage neuron to the dis-computing neuron, if the storage neuron is sigma_iTo dis computational neuron sigma_jPresence of directional synaptic connection, then d _ij1, otherwise d_ij＝0。

D₂＝(d_ij)_p×qIs a p × q-order matrix representing the directional synaptic connection relationship from the storage neuron to the max computation neuron, if the storage neuron is sigma_iCalculating neuron sigma by max_jPresence of directional synaptic connection, then d _ij1, otherwise d_ij＝0。

D₃＝(d_ij)_p×qIs of order p × qMatrix representing the directional synaptic connection relationship from the storage neuron to the min computation neuron, if the storage neuron is sigma_iCalculating neuron sigma by min_jPresence of directional synaptic connection, then d _ij1, otherwise d_ij＝0。

D₄＝(d_ij)_p×qIs a matrix of p × q order, which represents the directional synaptic connection relationship from the storage neuron to the rel computation neuron, if the storage neuron is sigma_iTo rel computing neuron sigma_jPresence of directional synaptic connection, then d _ij1, otherwise d_ij＝0。

E＝(e_ji)_q×pIs a q × order matrix representing the directional synaptic connection relationship from the computing neuron to the storage neuron, if the computing neuron is sigma_jTo the storage neuron σ_iExistence of synaptic connection, then e _ji1, otherwise e_ji＝0。

Δ represents dis calculation, and D^TΔθ＝(d₁,d₂,...,d_q) Wherein d is_j＝|d_1j×θ₁-d_2j×θ₂…-d_pj×θ_p|。

Denotes max calculation, and D^T·θ＝(d₁,d₂,...,d_q) Wherein d is_j＝max(d_1j×θ₁,d_2j×θ₂,...,d_pj×θ_p)。

Represents min calculation, and

wherein d is_j＝min(d_1j×θ₁,d_2j×θ₂...,d_pj×θ_p)。

Represents rel calculation, and

wherein

θ_maxRepresents the pulse value of the neuron, θ, located before the rel-computing neuron and after the max-computing neuron_disRepresenting the pulse value of the neuron before the rel calculating neuron and after the dis calculating neuron, wherein rho is a memory resolution coefficient and is in a fault diagnosis pi model_FThe median value is 0.5. Specially, when the fault diagnosis model II_FWhen there is only one memory event tag, i.e. there are only s sensing neurons and s memory neurons,

represents a tag-fetching calculation, an

Wherein

For tag-taking operations, if and only if

All internal non-0 elements being the same, d_j＝d_1j×τ₁Otherwise d_j＝0；

Wherein

If and only if

All internal non-0 elements being identical, e_i＝e_1i×ν₁Else, e_i＝0。

Represents a summation calculation, an

Wherein

The superscript T denotes the transpose of the vector sum matrix and the subscript g denotes the number of inference steps.

S45, calculating each fault diagnosis model pi_FThe average value f of the pulse values corresponding to the memory tag value_zAnd f is_zThe basic fault event represented by the memory label value corresponding to the maximum value is used as the fault diagnosis result of the suspected fault element, and the pulse value mean value f corresponding to the memory label value_zThe calculation formula of (2) is as follows:

wherein z represents the memory tag value of the output neuron, and the value is [0, c ]]The whole number of (a) to (b),

II representing positive sequence voltage fault diagnosis model_FPThe corresponding pulse value when the memory label value of the middle output neuron is z,

negative sequence voltage fault diagnosis model II_FNThe corresponding pulse value when the memory label value of the middle output neuron is z,

zero sequence voltage fault diagnosis model II_FZThe corresponding pulse value when the memory label value of the middle output neuron is z,

II model for representing remote signalling quantity fault diagnosis_FBThe corresponding pulse value when the memory tag value of the middle output neuron is 1,

II model for representing remote signalling quantity fault diagnosis_FBAnd when the memory tag value of the middle output neuron is 2, the corresponding pulse value is obtained.

In the following, the IEEE-14 node standard bus system is taken as a diagnosis target, and experimental example 1 is taken as an example, and a detailed diagnosis process of the present invention is given to facilitate detailed understanding.

Experimental example 1 the preset fault scenarios are shown in table 1:

TABLE 1

Firstly, when a fault occurs, the fault occurrence time is taken as a reference time (0s), and the telemetering quantity voltage amplitude values of 0.1s, 0.2s and 0.5s are taken respectively for normalization processing. Then, an FCM algorithm is used for cluster analysis to obtain a clustering result of the suspected fault area, as shown in fig. 12, where B12 is farthest from other elements, and is determined as the suspected fault area. Therefore, the bus B12 and the elements connected with the bus B12 are judged to be suspected fault elements, and a measuring tampering attack identification model II needs to be established for the suspected fault elements respectively_TNamely, it should be respectively applied to B12, L1206 and L1213 to establish a tamper attack identification model Π_TTake line L1213 as an example, obtain the pulse value curve of the sensing neuron and the memory neuron in the model, as shown in fig. 13.

As can be seen from FIG. 13, since the device L1213 in Experimental example 1 is not subjected to the tampering attack, the sensing neuron pulse value and the memory neuron pulse value are substantially matched, and then L1213 tamper attack recognition model II is performed_TTRMA algorithm to obtain a modelThe pulse values of the neurons are output as shown in fig. 14.

As can be seen from fig. 14, in the positive sequence (square), negative sequence (circular) and zero sequence (triangle) component measurement tampering attack recognition model of L1213, the output pulse values are all greater than the detection threshold 0.8, so there is no attack behavior at this time, and thus it is determined that L1213 is not attacked.

Finally, a line L1213 fault diagnosis model pi is established_FAnd executing the FDMA algorithm to obtain the memory tag value of the model output neuron and the corresponding pulse value thereof, as shown in Table 2.

TABLE 2

As can be seen from table 2, when the label value is 2, the pulse values of the output neurons of the positive sequence component diagnosis model, the negative sequence component diagnosis model and the zero sequence component diagnosis model of L1213 all exceed 0.95, and in addition, the fault diagnosis model Π_FWhen the tag value is 2, f₂0.979, which is significantly larger than other label values, and therefore the fault event corresponding to the label value 2 is determined as the result of the fault diagnosis, that is, the single-phase ground fault occurs on the line L1213, and the fault reliability is 0.979.

The advantageous effects of the present invention are further illustrated by the following experimental example 2:

experimental example 2 the preset fault scenarios are shown in table 3:

TABLE 3

Fig. 15 shows the result of the fault diagnosis method shown in the embodiment of the present invention, which first detects a tamper attack, the abscissa in fig. 15 is the neuron number, and the ordinate is the pulse value size, it is obvious that when the neuron number is greater than 1600 in fig. 15(a), the pulse value (portion indicated by triangle) output by the zero sequence diagnosis model is significantly lower than the detection threshold 0.8, and also the pulse value (portion indicated by circle) output by the negative sequence diagnosis model is partially lower than 0.8, according to the detection determination method of the embodiment of the present invention, it is considered that the tamper attack occurs at this time, the element L0102 is attacked by the tamper attack, while in fig. 15(B), the pulse values of the positive sequence diagnosis model (portion indicated by square), the negative sequence diagnosis model (portion indicated by circle) and the zero sequence diagnosis model (portion indicated by triangle) output neurons are almost distributed over 0.96 and much larger than the detection threshold 0.8, according to the attack identification method of the embodiment of the present invention, the element B1 is not attacked by the measurement attack, which is consistent with the preset fault, in this case, in this situation, the SCADA system does not detect the fault diagnosis method, and this situation is shown in this false diagnosis scenario 4.

TABLE 4

In table 4, M at the upper left corner of the faulty element indicates that the element has fault diagnosis, and it can be seen that in the case of experimental example 2, since other conventional fault diagnosis methods do not have the capability of identifying the tamper attack, when the tamper attack occurs, misdiagnosis of L0102 elements occurs in these methods, but the method of the present invention establishes an effective tamper attack identification model, and can effectively identify the attacked behavior of the L0102 element in experimental example 2, in addition, the experimental example 2 also has the remote traffic error information related to the bus B2, at this time, the conventional method based on remote traffic may have fault diagnosis for the B2 element, but the method of the present invention does not have fault diagnosis for the B2, because the present invention comprehensively utilizes the remote traffic and the remote traffic information to implement fault diagnosis, which can reduce the diagnosis risk based on single fault information.

TABLE 5

Table 5 shows the comparison of the modeling performance of each method, and it can be seen that the method of the present invention comprehensively utilizes the fault telemetry and telemetry, which can obtain richer diagnosis results than other methods. The fault diagnosis method does not need the complicated and huge learning and training process of the ANN method, so that the timeliness is better; the method does not need a modeling mode based on the production rule similar to FPNS, tFRSNPS and IFSNP, so that the fault remote measurement and remote signaling quantity can be effectively and comprehensively utilized, the diagnosis result is richer, and the specific fault type can be diagnosed.

It will be appreciated by those of ordinary skill in the art that the embodiments described herein are intended to assist the reader in understanding the principles of the invention and are to be construed as being without limitation to such specifically recited embodiments and examples. Those skilled in the art can make various other specific changes and combinations based on the teachings of the present invention without departing from the spirit of the invention, and these changes and combinations are within the scope of the invention.

Claims (10)

1. The smart grid fault diagnosis method capable of identifying and measuring the tampering attack is characterized by comprising the following steps of:

s1, aiming at elements which have faults in the smart grid and have completed the breaker on-off detection, determining suspected fault elements by adopting an electric quantity clustering detection method;

s2, respectively establishing a measuring tampering attack recognition model based on a memory pulse neurolemma system for each suspected fault element, solving the model through a TRMA algorithm, judging whether each suspected fault element suffers measuring tampering attack, if so, entering a step S3, and if not, entering a step S4;

s3, judging that the action of the relay protection device at the moment is a false action caused by measuring tampering attack telemeasurement, and finishing fault diagnosis of the smart power grid;

s4, aiming at each suspected fault element which is not subjected to the measurement tampering attack, respectively establishing a fault diagnosis model based on a memory pulse neural membrane system according to the topological structure and the protection pattern of the power grid, solving through an FDMA algorithm to obtain a fault diagnosis result of the suspected fault element, and finishing the fault diagnosis of the intelligent power grid.

2. The smart grid fault diagnosis method according to claim 1, wherein the step S1 includes the following sub-steps:

s11, when a relay in the intelligent power grid acts or a breaker trips, acquiring voltage values in each bus, generator and transformer in 0.1S, 0.2S and 0.5S in the intelligent power grid;

s12, performing cluster analysis on the normalized voltage value and the normalized voltage value of the smart grid in the normal operation state, wherein the normalization formula is as follows:

wherein x' represents the normalized voltage value, x represents the voltage value to be normalized, and x_maxAnd x_minRespectively a maximum value and a minimum value in the voltage values to be normalized;

and S13, taking the element farthest from the cluster center as a suspected fault element.

3. The smart grid fault diagnosis method according to claim 1, wherein the step S2 includes the following sub-steps:

s21, respectively establishing a tamper attack detection model pi based on a memory pulse neurolemma system for each suspected fault element_T；

S22, carrying out normalization processing on SCADA remote measurement and RTU remote measurement of the suspected fault element at the moment of fault to obtain a real-time fault voltage normalization value of SCADA voltage and RTU voltage within a sampling time range;

s23, real-time fault electricity of the SCADA voltage and the RTU voltage in a sampling time rangeRespectively splitting the voltage normalization value into s sampling time points, and inputting the SCADA real-time fault voltage normalization value of each sampling time point into a measuring tampering attack identification model pi_TIn the perception neuron, the RTU real-time fault voltage normalization value of each sampling time point is input into a measurement tampering attack recognition model pi_TIn the memory neurons of (a);

s24, identifying pi through TRMA algorithm for measuring and tampering attack of each suspected fault element_TSolving to obtain various measurement tampering attack identification models pi_TThe pulse value of the neuron is output;

s25, sequentially judging pi of each measured tampering attack recognition model_TIf the pulse value of any output neuron is lower than the detection threshold value, the suspected faulty element is determined to be attacked by measurement tampering, and the step S3 is entered, otherwise, the suspected faulty element is determined not to be attacked by measurement tampering, and the step S4 is entered.

4. The smart grid fault diagnosis method according to claim 3, wherein the measured tampering attack recognition model Π based on the memory pulse neurolemma system established in step S21_TThe method specifically comprises the following steps:

Π_T＝(O,σ₁,...,σ_m,syn,in,out)

where O ═ { a } represents a set of nerve pulses, a represents a nerve pulse, σ represents a set of nerve pulses, and₁,...,σ_midentifying model Π for measuring tampering attacks_TM neurons of (1), σ_i＝(θ_i,λ_i,τ_i,r_i),i＝1,2,...,m，θ_iThe pulse value of the ith neuron is represented by [0,1 ]]The real number of (c); lambda [ alpha ]_iExpressing the ignition threshold value of the ith neuron, wherein the value of the ignition threshold value is any real number; tau is_iThe memory tag value of the ith neuron is [0, c ]]C is the total number of memory events; r is_iIndicating the firing rule of the ith neuron, in the form of E/a^(θ,τ)→a^(β,τ)Wherein E ═ aⁿ,θ≥λ_iIs the ignition condition, meaning if and only if godChannel element sigma_iReceiving at least n pulses and the neuron pulse value satisfies theta ≧ lambda_iThe firing rule can only be executed, at which time the neuron σ_iA pulse with potential value theta is consumed, a new pulse with potential value β is generated and transmitted backwards, and at the same time, the label value tau of the neuron_iIs consumed and passes a new tag value tau backward_iOtherwise, the neuron does not execute ignition calculation, syn {1, 2.. multidot.m } × {1, 2.. multidot.m } represents directed synaptic connection relation between neurons, and in and out represent respectively a measuring and tampering attack recognition model pi_TA set of input neurons and a set of output neurons.

5. The smart grid fault diagnosis method according to claim 4, wherein the measured tampering attack recognition model Π_TTamper attack identification model pi including positive sequence voltage measurement_TPNegative sequence voltage measurement tampering attack identification model II_TNTampering attack recognition model pi with zero sequence voltage measurement_TZ；

II (II) of measurement tampering attack recognition model_TThe neurons in (1) comprise storage neurons for sensing or storing the pulse values and the memory event tag values from the environment and computation neurons for computing the pulse values and the memory event tag values;

the storage neurons comprise perception neurons, memory neurons and transmission neurons; the perception neuron is used for acquiring detection information from the environment, converting the detection information into real-time basic event characteristics and transmitting the real-time basic event characteristics backwards in the form of nerve pulse values; the memory neuron is used for storing memory basic event characteristics and corresponding memory tag values; the transmission neuron is used for transmitting a pulse value and a memory tag value transmitted by the pre-synaptic neuron to the post-synaptic neuron;

the calculating neurons comprise dis calculating neurons, max calculating neurons, min calculating neurons and rel calculating neurons; the dis calculation neuron is used for acquiring the matching degree between the current SCADA real-time basic fault characteristics and the RTU real-time basic fault characteristics by calculating the difference degree of pulse values of the perception neuron and the memory neuron; the max calculating neuron is used for carrying out logic OR calculation on the impulse value transmitted by the pre-synaptic neuron, namely taking the maximum impulse value of the pre-synaptic neuron as the impulse value of the current max calculating neuron to participate in subsequent calculation; the min calculation neuron is used for carrying out logic AND calculation on the pulse value transmitted by the pre-synaptic neuron, namely taking the minimum pulse value of the pre-synaptic neuron as the pulse value of the current min calculation neuron to participate in subsequent calculation; the rel computing neuron is used for carrying out global matching degree computation on pulse values transmitted by the pre-synaptic neurons, namely finding out the matching degree of the transmission neuron with the minimum number in the pre-synaptic neurons in the global state.

6. The smart grid fault diagnosis method according to claim 5, wherein in the step S24, the measured tampering attack recognition model II for each suspected fault element through the TRMA algorithm is adopted_TThe specific method for solving is as follows:

a1, setting the inference step number g to be 0;

a2, performing ignition calculation for each storage neuron meeting the ignition condition, and updating according to the following formula_g+1And v_g+1：

A3, performing ignition calculation for each calculation neuron meeting the ignition condition, and updating theta according to the following formula_g+1And τ_g+1：

A4, adding 1 to the inference step number g;

a5, judging whether the operation condition theta is met_g≠0₁Or_g≠0₂If yes, returning to the step A2, otherwise, ending the TRMA algorithm, and outputting the measured tampering attack recognition model II_TThe pulse value of the neuron is output;

the meaning of the vector, the matrix and the operator involved in the TRMA algorithm is as follows:

θ＝(θ₁,θ₂,...,θ_p)^Trepresenting a vector of pulse values of a storage neuron, where θ_iThe pulse value of the ith storage neuron is [0,1 ]]The real number above, i 1,2, p, p denotes the number of storage neurons; tampering attack identification model II in the measurement_TIn the case where 1. ltoreq. i. ltoreq.s, θ_iFor the SCADA real-time fault voltage normalization value, when (s +1) is less than or equal to i and less than or equal to (s + s c), theta_iThe normalized value is the RTU real-time fault voltage, c is the total number of memory events, and s is the total number of sampling time points;

＝(₁,₂,...,_q)^Trepresenting a vector of calculated neuron pulse values, wherein_jCalculating the impulse value of the neuron for the jth, and taking the value as [0,1 ]]The real number above, j 1,2,.., q, q represents the number of computational neurons;

τ＝(τ₁,τ₂,...,τ_p)^Trepresenting a vector of storage neuron tag values, where τ_iThe tag value for the ith storage neuron is set to [0, c]The above integer; tampering attack identification model II in the measurement_TIn (1) i.ltoreq (s + s c), τ is_iAll are 0;

ν＝(ν₁,ν₂,...,ν_q)^Trepresenting a vector of computed neuron tag values, where v_jCalculating the label value of the neuron for the jth, and taking the value as [0, c]The above integer;

D₁＝(d_ij)_p×qis a p × q-order matrix representing the directional synaptic connection relationship from the storage neuron to the dis-computing neuron, if the storage neuron is sigma_iTo dis computational neuron sigma_jPresence of directional synaptic connection, then d_ij1, otherwise d_ij＝0；

D₂＝(d_ij)_p×qIs a p × q-order matrix representing the directional synaptic connection relationship from the storage neuron to the max computation neuron, if the storage neuron is sigma_iCalculating neuron sigma by max_jPresence of directional synaptic connection, then d_ij1, otherwise d_ij＝0；

D₃＝(d_ij)_p×qIs a matrix of p × q order, which represents the directional synaptic connection relationship from the storage neuron to the min computation neuron, if the storage neuron is sigma_iCalculating neuron sigma by min_jPresence of directional synaptic connection, then d_ij1, otherwise d_ij＝0；

D₄＝(d_ij)_p×qIs a matrix of p × q order, which represents the directional synaptic connection relationship from the storage neuron to the rel computation neuron, if the storage neuron is sigma_iTo rel computing neuron sigma_jPresence of directional synaptic connection, then d_ij1, otherwise d_ij＝0；

E＝(e_ji)_q×pIs a q × order matrix representing the directional synaptic connection relationship from the computing neuron to the storage neuron, if the computing neuron is sigma_jTo the storage neuron σ_iExistence of synaptic connection, then e_ji1, otherwise e_ji＝0；

Δ represents dis calculation, and D^TΔθ＝(d₁,d₂,...,d_q) Wherein d is_j＝|d_1j×θ₁-d_2j×θ₂...-d_pj×θ_p|；

Denotes max calculation, and D^T·θ＝(d₁,d₂,...,d_q) Wherein d is_j＝max(d_1j×θ₁,d_2j×θ₂,...,d_pj×θ_p)；

Represents min calculation, and

wherein d is_j＝min(d_1j×θ₁,d_2j×θ₂...,d_pj×θ_p)；

Represents rel calculation, and

wherein

θ_maxRepresents the pulse value of the neuron, θ, located before the rel-computing neuron and after the max-computing neuron_disRepresenting the pulse value of the neuron before the rel computing neuron and after the dis computing neuron, wherein rho is a memory resolution coefficient, and the model II is identified in the measurement tampering attack recognition model_TThe median value is 1;

represents a tag-fetching calculation, an

Wherein

For tag-taking operations, if and only if

All internal non-0 elements being the same, d_j＝d_1j×τ₁Otherwise d_j＝0；

Wherein

If and only if

All internal non-0 elements being identical, e_i＝e_1i×ν₁Else, e_i＝0；

Represents a summation calculation, an

Wherein

The superscript T denotes the transpose of the vector sum matrix and the subscript g denotes the number of inference steps.

7. The smart grid fault diagnosis method according to claim 1, wherein the step S4 includes the following sub-steps:

s41, respectively establishing a fault diagnosis model pi based on a memory pulse neurolemma system according to the topological structure and the protection pattern of the power grid aiming at each suspected fault element which is not subjected to the measurement tampering attack_F(ii) a II fault diagnosis model_FII comprises a positive sequence voltage fault diagnosis model_FPNegative sequence voltage fault diagnosis model II_FNZero sequence voltage fault diagnosis model II_FZII model for diagnosing sum remote signaling quantity fault_FB；

S42, respectively carrying out normalization processing on the SCADA telemetering value of the suspected fault element and the historical telemetering values of different types of faults of the element to obtain an SCADA real-time fault voltage normalization value and an SCADA historical fault voltage normalization value;

s43, dividing the SCADA real-time fault voltage normalization value and the SCADA historical fault voltage normalization value into S sampling time points respectively, and inputting the SCADA real-time fault voltage normalization value of each sampling time point into the positive sequence voltage fault diagnosis model pi respectively_FPNegative sequence voltage fault diagnosis model II_FNII model for diagnosing fault of sum zero sequence voltage_FZIn the sensing neuron, a remote traffic pulse value related to the element is input into a remote traffic fault diagnosis model pi_FBIn the perception neuron, the normalized values of the SCADA historical fault voltage at each sampling time point are respectively input into a positive sequence voltage fault diagnosis model II_FPNegative sequence voltage fault diagnosis model II_FNII model for diagnosing fault of sum zero sequence voltage_FZIn the memory neuron, a history fault remote communication quantity related to the element is inputted as a pulse value to a remote communication quantity fault diagnosis model II_FBIn the memory neurons of (a);

s44, carrying out fault diagnosis model pi on each suspected fault element through FDMA algorithm_FSolving to obtain various fault diagnosis models pi_FThe pulse value and the memory tag value of the neuron are output;

s45, calculating each fault diagnosis model pi_FThe average value f of the pulse values corresponding to the memory tag value_zAnd f is_zThe basic fault event represented by the memory label value corresponding to the maximum value is used as the fault diagnosis result of the suspected fault element, and the pulse value mean value f corresponding to the memory label value_zThe calculation formula of (2) is as follows:

wherein z represents the memory tag value of the output neuron, and the value is [0, c ]]The whole number of (a) to (b),

indicating positive sequence electricityPressure fault diagnosis model pi_FPThe corresponding pulse value when the memory label value of the middle output neuron is z,

negative sequence voltage fault diagnosis model II_FNThe corresponding pulse value when the memory label value of the middle output neuron is z,

zero sequence voltage fault diagnosis model II_FZThe corresponding pulse value when the memory label value of the middle output neuron is z,

II model for representing remote signalling quantity fault diagnosis_FBThe corresponding pulse value when the memory tag value of the middle output neuron is 1,

II model for representing remote signalling quantity fault diagnosis_FBAnd when the memory tag value of the middle output neuron is 2, the corresponding pulse value is obtained.

8. The smart grid fault diagnosis method according to claim 7, wherein the fault diagnosis model Π based on the memory pulse neurolemma system established in step S41_FThe method specifically comprises the following steps:

Π_F＝(O,σ₁,...,σ_m,syn,in,out)

where O ═ { a } represents a set of nerve pulses, a represents a nerve pulse, σ represents a set of nerve pulses, and₁,...,σ_mmodel pi for fault diagnosis_FM neurons of (1), σ_i＝(θ_i,λ_i,τ_i,r_i),i＝1,2,...,m，θ_iThe pulse value of the ith neuron is represented by [0,1 ]]The real number of (c); lambda [ alpha ]_iExpressing the ignition threshold value of the ith neuron, wherein the value of the ignition threshold value is any real number; tau is_iThe memory tag value of the ith neuron is [0, c ]]C is an integer ofThe total number of the memory events; r is_iIndicating the firing rule of the ith neuron, in the form of E/a^(θ,τ)→a^(β,τ)Wherein E ═ aⁿ,θ≥λ_iIs the firing condition, meaning if and only if the neuron σ_iReceiving at least n pulses and the neuron pulse value satisfies theta ≧ lambda_iThe firing rule can only be executed, at which time the neuron σ_iA pulse with potential value theta is consumed, a new pulse with potential value β is generated and transmitted backwards, and at the same time, the label value tau of the neuron_iIs consumed and passes a new tag value tau backward_iOtherwise, the neuron does not execute ignition calculation, syn {1, 2.. multidot.m } × {1, 2.. multidot.m } represents directed synaptic connection relation between neurons, and in and out represent fault diagnosis models Π and Π respectively_FA set of input neurons and a set of output neurons.

9. The smart grid fault diagnosis method according to claim 8, wherein the fault diagnosis model Π is_FThe neurons in (1) comprise storage neurons for sensing or storing the pulse values and the memory event tag values from the environment and computation neurons for computing the pulse values and the memory event tag values;

the storage neurons comprise perception neurons, memory neurons and transmission neurons; the perception neuron is used for acquiring detection information from the environment, converting the detection information into real-time basic event characteristics and transmitting the real-time basic event characteristics backwards in the form of nerve pulse values; the memory neuron is used for storing memory basic event characteristics and corresponding memory tag values; the transmission neuron is used for transmitting a pulse value and a memory tag value transmitted by the pre-synaptic neuron to the post-synaptic neuron;

the calculating neurons comprise dis calculating neurons, max calculating neurons, min calculating neurons and rel calculating neurons; the dis calculation neuron is used for acquiring the matching degree between the current real-time basic fault feature and the historical fault feature by calculating the difference degree of pulse values of the perception neuron and the memory neuron; the max calculating neuron is used for carrying out logic OR calculation on the impulse value transmitted by the pre-synaptic neuron, namely taking the maximum impulse value of the pre-synaptic neuron as the impulse value of the current max calculating neuron to participate in subsequent calculation; the min calculation neuron is used for carrying out logic AND calculation on the pulse value transmitted by the pre-synaptic neuron, namely taking the minimum pulse value of the pre-synaptic neuron as the pulse value of the current min calculation neuron to participate in subsequent calculation; the rel computing neuron is used for carrying out global matching degree computation on pulse values transmitted by the pre-synaptic neurons, namely finding out the matching degree of the transmission neuron with the minimum number in the pre-synaptic neurons in the global state.

10. The smart grid fault diagnosis method according to claim 9, wherein the fault diagnosis model Π for each suspected fault element in step S44 is implemented through FDMA algorithm_FThe specific method for solving is as follows:

b1, setting the inference step number g to be 0;

b2, performing ignition calculation on each storage neuron meeting the ignition condition, and updating according to the following formula_g+1And v_g+1：

B3, performing ignition calculation on each calculation neuron meeting the ignition condition, and updating theta according to the following formula_g+1And τ_g+1：

B4, adding 1 to the inference step number g;

b5, judging whether the operation condition theta is met_g≠0₁Or_g≠0₂If yes, returning to the step A2, otherwise, ending the FDMA algorithm, and outputting to obtain each fault diagnosis model II_FThe pulse value and the memory tag value of the neuron are output;

the vector, matrix and operator involved in the FDMA algorithm have the following meanings:

θ＝(θ₁,θ₂,...,θ_p)^Trepresenting a vector of pulse values of a storage neuron, where θ_iThe pulse value of the ith storage neuron is [0,1 ]]The real number above, i 1,2, p, p denotes the number of storage neurons; II in the fault diagnosis model_FIn the case where 1. ltoreq. i. ltoreq.s, θ_iFor the SCADA real-time fault voltage normalization value, when (s +1) is less than or equal to i and less than or equal to (s + s c), theta_iThe method comprises the steps that the historical fault voltage normalization values of SCADA of different fault types are obtained, c is the total number of memory events, and s is the total number of sampling time points;

＝(₁,₂,...,_q)^Trepresenting a vector of calculated neuron pulse values, wherein_jCalculating the impulse value of the neuron for the jth, and taking the value as [0,1 ]]The real number above, j 1,2,.., q, q represents the number of computational neurons;

τ＝(τ₁,τ₂,...,τ_p)^Trepresenting a vector of storage neuron tag values, where τ_iThe tag value for the ith storage neuron is set to [0, c]The above integer; II in the fault diagnosis model_FIn the case where 1. ltoreq. i.ltoreq.s, τ_iAll are 0, and when (s +1) ≦ i ≦ (s + s × c), τ is_iTo remember the basic fault event type;

ν＝(ν₁,ν₂,...,ν_q)^Trepresenting a vector of computed neuron tag values, where v_jCalculating the label value of the neuron for the jth, and taking the value as [0, c]The above integer;

D₁＝(d_ij)_p×qis a p × q-order matrix representing the directional synaptic connection relationship from the storage neuron to the dis-computing neuron, if the storage neuron is sigma_iTo dis meterCalculating neuron sigma_jPresence of directional synaptic connection, then d_ij1, otherwise d_ij＝0；

D₂＝(d_ij)_p×qIs a p × q-order matrix representing the directional synaptic connection relationship from the storage neuron to the max computation neuron, if the storage neuron is sigma_iCalculating neuron sigma by max_jPresence of directional synaptic connection, then d_ij1, otherwise d_ij＝0；

D₃＝(d_ij)_p×qIs a matrix of p × q order, which represents the directional synaptic connection relationship from the storage neuron to the min computation neuron, if the storage neuron is sigma_iCalculating neuron sigma by min_jPresence of directional synaptic connection, then d_ij1, otherwise d_ij＝0；

D₄＝(d_ij)_p×qIs a matrix of p × q order, which represents the directional synaptic connection relationship from the storage neuron to the rel computation neuron, if the storage neuron is sigma_iTo rel computing neuron sigma_jPresence of directional synaptic connection, then d_ij1, otherwise d_ij＝0；

E＝(e_ji)_q×pIs a q × order matrix representing the directional synaptic connection relationship from the computing neuron to the storage neuron, if the computing neuron is sigma_jTo the storage neuron σ_iExistence of synaptic connection, then e_ji1, otherwise e_ji＝0；

Δ represents dis calculation, and D^TΔθ＝(d₁,d₂,...,d_q) Wherein d is_j＝|d_1j×θ₁-d_2j×θ₂...-d_pj×θ_p|；

Denotes max calculation, and D^T·θ＝(d₁,d₂,...,d_q) Wherein d is_j＝max(d_1j×θ₁,d_2j×θ₂,...,d_pj×θ_p)；

Represents min calculation, and

wherein d is_j＝min(d_1j×θ₁,d_2j×θ₂...,d_pj×θ_p)；

Represents rel calculation, and

wherein

θ_maxRepresents the pulse value of the neuron, θ, located before the rel-computing neuron and after the max-computing neuron_disRepresenting the pulse value of the neuron before the rel computing neuron and after the dis computing neuron, wherein rho is a memory resolution coefficient, and pi is a fault diagnosis model_FThe median value is 0.5;

represents a tag-fetching calculation, an

It is composed of

For tag-taking operations, if and only if

Each internal non-0 element is the sameWhen d is greater than_j＝d_1j×τ₁Otherwise d_j＝0；

Wherein

If and only if

All internal non-0 elements being identical, e_i＝e_1i×ν₁Else, e_i＝0；

Represents a summation calculation, an

Wherein

The superscript T denotes the transpose of the vector sum matrix and the subscript g denotes the number of inference steps.

Images