CN113505458A - Cascading failure key trigger branch prediction method, system, equipment and storage medium - Google Patents

Cascading failure key trigger branch prediction method, system, equipment and storage medium Download PDF

Info

Publication number
CN113505458A
CN113505458A CN202110846723.3A CN202110846723A CN113505458A CN 113505458 A CN113505458 A CN 113505458A CN 202110846723 A CN202110846723 A CN 202110846723A CN 113505458 A CN113505458 A CN 113505458A
Authority
CN
China
Prior art keywords
branch
cascading failure
power
cascading
predicted
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202110846723.3A
Other languages
Chinese (zh)
Other versions
CN113505458B (en
Inventor
乔骥
古思丽
张东霞
王晓辉
王新迎
任汉涛
陈二松
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
State Grid Corp of China SGCC
Electric Power Research Institute of State Grid Hebei Electric Power Co Ltd
China Electric Power Research Institute Co Ltd CEPRI
Original Assignee
State Grid Corp of China SGCC
Electric Power Research Institute of State Grid Hebei Electric Power Co Ltd
China Electric Power Research Institute Co Ltd CEPRI
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by State Grid Corp of China SGCC, Electric Power Research Institute of State Grid Hebei Electric Power Co Ltd, China Electric Power Research Institute Co Ltd CEPRI filed Critical State Grid Corp of China SGCC
Priority to CN202110846723.3A priority Critical patent/CN113505458B/en
Publication of CN113505458A publication Critical patent/CN113505458A/en
Application granted granted Critical
Publication of CN113505458B publication Critical patent/CN113505458B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/10—Geometric CAD
    • G06F30/18—Network design, e.g. design based on topological or interconnect aspects of utility systems, piping, heating ventilation air conditioning [HVAC] or cabling
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00—Details relating to CAD techniques
    • G06F2111/08—Probabilistic or stochastic CAD
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/06—Power analysis or power optimisation
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
    • Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
    • Y04S10/00—Systems supporting electrical power generation, transmission or distribution
    • Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Geometry (AREA)
  • Artificial Intelligence (AREA)
  • Software Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computer Hardware Design (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Computational Mathematics (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Pure & Applied Mathematics (AREA)
  • Supply And Distribution Of Alternating Current (AREA)

Abstract

The invention provides a prediction method, a system, equipment and a storage medium for a critical trigger branch of cascading failures, wherein the method comprises the following steps: acquiring online prediction target information; constructing online prediction target information into a graph data sample set, and integrating the trend and the topological state into a branch characteristic matrix and an incidence matrix; and inputting the branch characteristic matrix and the incidence matrix into a cascading failure simulation model to identify weak branches which possibly cause serious cascading failures, and obtaining the weak degree of each branch in the power system to be predicted. The method obviously improves the working efficiency of preventing the cascading failure, and has greater engineering application value and popularization prospect for preventing and controlling the cascading failure.

Description

Cascading failure key trigger branch prediction method, system, equipment and storage medium
Technical Field
The invention belongs to the field of power grid weak link identification and analysis, and particularly relates to a cascading failure key trigger branch prediction method, a system, equipment and a storage medium.
Background
Cascading failure blackouts can be summarized as the process of power flow diversion caused by the initial element failure, causing subsequent elements to cascade out of operation and ultimately causing a system blackout. Because the development and the updating of the power system lead the influence factors of the cascading failure to be increasingly complex, and the consequences are serious and unavoidable once the cascading failure occurs, the identification and the weak factor analysis of the key trigger branch of the cascading failure are carried out based on the real-time state of the power grid, so that the method has important practical significance for the advanced prediction and the prevention control of the cascading failure, the large power failure accident risk reduction and the safety and the stability of the power system. Because the safety state of the power grid is determined by the topological structure of the element and the operation state of the element together, the weak degree of the branch and the structural position and the state of the branch have a large relationship, the limitation of calculating the weak branch by the conventional index needs to be broken, and the real-time operation state is analyzed. Due to the complex characteristic of cascading failures, how to comprehensively combine the running state with the topological structure, and how to quickly and accurately predict weak branches which may trigger serious cascading failures is a difficult point.
Disclosure of Invention
In order to solve the problem of weak branch prediction of cascading failures in the prior art, the invention provides a method, a system, equipment and a storage medium for predicting a key trigger branch of the cascading failures.
In order to achieve the purpose, the invention adopts the following technical scheme:
a cascading failure key triggering branch prediction method comprises the following steps:
acquiring online prediction target information;
constructing online prediction target information into a graph data sample set, and integrating the trend and the topological state into a branch characteristic matrix and an incidence matrix;
and inputting the branch characteristic matrix and the incidence matrix into a cascading failure simulation model to identify weak branches which possibly cause serious cascading failures, and obtaining the weak degree of each branch in the power system to be predicted.
As a further improvement of the present invention, the online prediction target information includes one or more of a complete topology of the power system to be predicted, parameters of each element, injected active power, injected reactive power, active load, reactive load of each node, and a voltage and a phase angle of each node.
As a further improvement of the present invention, the building process of the cascading failure simulation model includes:
acquiring reference training target information, wherein the target information comprises a complete topological structure of a target network and basic power flow state information;
generating cascading failure accident chain data by the reference training target information; arranging cascading failure accident chain data, taking the information of the trend and the topological state as input, taking the branch weak index of each branch in the state as output, and generating a training sample set;
and constructing and training a cascading failure simulation model based on the depth map convolutional neural network.
As a further improvement of the invention, the generating of the cascading failure accident chain data from the reference training target information is to generate the cascading failure accident chain data by modifying the power grid topological structure and the tidal current state and performing the cascading failure simulation.
As a further improvement of the present invention, the generating the cascading failure accident chain data specifically includes:
aiming at the branch topology change condition, constructing a new topology structure by randomly breaking one or two branches;
under the constructed topological structure, aiming at the situations of load change, generator output regulation and reactive device switching in an actual power grid, injecting active power and reactive power into each node, and carrying out random graded fluctuation regulation on the active load and the reactive load to generate and record an initial tide running state;
under the initial tide running state, selecting a certain branch as a cascading failure triggering branch, and disconnecting the branch;
carrying out load shedding operation on the node when the load shedding operation is carried out on the node when the node voltage is lower than the safety threshold value, wherein the load shedding amount is determined by a dichotomy; calculating branch load flow to calculate the branch outage probability, and disconnecting the branch when the branch outage probability is found to be greater than a threshold value; recording fault chain data until no branch outage probability is greater than a threshold value and the node voltage is lower than a safety threshold value;
traversing all branches of the power grid as cascading failure triggering branches to obtain cascading failure chain data triggered by each branch of the whole power grid in an initial tide running state;
and then cascading failure chain data under various topological structures and tidal current states are constructed and recorded.
As a further improvement of the invention, the branch outage probability in the power flow calculation of the power system is obtained by the following method:
Figure BDA0003180862140000031
in the formula, Pl、PmaxK is the transmission power of the branch, the upper limit of the transmission power of the branch, the limit transmission power multiple, pwFor branch protection stealth failure probability, pcThe branch outage probability when the power reaches the limit power.
As a further improvement of the present invention, the sorting of the cascading failure accident chain data specifically includes:
the cascading failure accident chain data are constructed into a graph data sample set, the branch weak index of a branch is constructed by adopting the cascading failure chain loss triggered by the branch in a certain state, the load flow and the topological state are integrated to form a training sample, and the training sample comprises a branch characteristic matrix, an incidence matrix and a branch weak index.
As a further improvement of the invention, the branch characteristic matrix XGCNIICovering the tidal current state information, expressed as:
XGCNII＝[xrelay,xp,xratio,xl,xloss,xfv,xftheta,xfpgen,xtv,xttheta,xtpload]
wherein x isrelay、xp、xratioThe branch transmission power, the limit transmission power and the ratio of the two represent the influence of overload protection and hidden fault xl、xlossRespectively defining a source end node, a tail end node and x according to the branch power flow direction for branch electric sodium and active lossfv,xftheta,xfpgenInjecting active power for voltage amplitude, voltage phase angle and power generation of a branch source end node respectively; x is the number oftv,xttheta,xtploadRespectively outputting active power for the voltage amplitude, the voltage phase angle and the load of the tail end node of the branch circuit;
incidence matrix AGCNIITopology information is covered, represented as an N0-1 matrix:
Figure BDA0003180862140000041
branch weakness index YGCNII∈RN×kDividing the weak degree of the N branches into k types from small to large:
Figure BDA0003180862140000042
as a further improvement of the method, interpretable analysis is carried out on the weak degree of each branch in the power system to be predicted, and weak cause of the predicted branch is obtained through solving; the method specifically comprises the following steps:
selecting a mapping sample (Y)GCNII,XGCNII,AGCNII) Using GNN Expiainer algorithm as prediction result
Figure BDA0003180862140000049
Generating a subset of interpretations
Figure BDA0003180862140000043
Wherein
Figure BDA0003180862140000044
Respectively represent the pair predictions
Figure BDA0003180862140000045
Neighborhood branch sets and feature subsets with significant impact;
the GNN Explainer algorithm optimizes the structure mask σ (M) and feature mask σ (F) of the entire computation graph, respectively, with the goal of mutual information maximization based on the mutual information model MI:
Figure BDA00031808621400000410
where Y is the distribution of the predicted values for a branch, H (g) represents the entropy calculation for a discrete quantity, M ∈ RN×NFor the computed sub-graph mask parameters to be learned, AcIs a adjacency matrix, C represents the number of classes of the classification problem, σ (g) is a sigmond function, XS＝{xjvj∈GSIndicates neighborhood branchCollection GSAll branch characteristics;
optimizing to obtain mutual information with the predicted value of the current model, extracting and predicting the value
Figure BDA0003180862140000046
Graph domain interpretation subset with highest mutual information
Figure BDA0003180862140000047
Thereby obtaining the weak cause of the predicted branch
Figure BDA0003180862140000048
A cascading failure critical triggering branch prediction system, comprising:
the information acquisition module is used for acquiring online prediction target information comprising a complete topological structure of a target network and basic power flow state information;
the data construction module is used for arranging the online prediction target information into a branch characteristic matrix and an incidence matrix so as to construct prediction input data;
and the branch prediction module is used for identifying weak branches which possibly cause serious cascading failure from the prediction input data through the cascading failure simulation model to obtain the weak degree of each branch in the power system to be predicted.
The system also comprises a cause analysis module which is used for performing interpretable analysis on the weakness degree of each branch in the power system to be predicted and solving to obtain the weakness cause of the predicted branch.
An electronic device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing the steps of the cascading failure critical triggering branch prediction method when executing the computer program.
A computer readable storage medium storing a computer program which, when executed by a processor, implements the steps of the cascading failure critical triggering branch prediction method.
Compared with the prior art, the invention has the beneficial effects that:
according to the method, online prediction target information is sorted into a branch characteristic matrix and an incidence matrix according to a trained cascading failure simulation model and a certain power grid topological structure and an operation state so as to construct prediction input data; and identifying weak branches which possibly cause serious cascading failures by the predicted input data through a cascading failure simulation model to obtain the weak degree of each branch in the power system to be predicted. The method obviously improves the working efficiency of preventing the cascading failure, and has greater engineering application value and popularization prospect for preventing and controlling the cascading failure.
Furthermore, the depth map convolutional neural network is constructed and trained on the basis of the depth map convolutional neural network, weak branches which may trigger serious cascading failures can be predicted on the basis of real-time power flow and topological state, explanation analysis is carried out on weak branch cause by adopting a map neural network explanation algorithm, neighborhood elements and characteristics which have obvious influence on the weak branches are excavated, the working efficiency of preventing cascading failures is obviously improved, and the method has a great engineering application value and a popularization prospect in preventing and controlling the cascading failures.
Furthermore, the grid topological structure information can be learned by adopting a depth map convolution neural network, online application is carried out aiming at different topological structures, and the method for predicting the cascading failure key trigger branch and analyzing the weak cause of the interpretable algorithm has good generalization performance and interpretability.
Drawings
FIG. 1 is a schematic flow chart of a method for predicting a critical triggering branch of a cascading failure according to the present invention;
FIG. 2 is a flow chart of a cascading failure key triggering branch prediction and weakness cause analysis method based on a depth map convolutional neural network and an interpretable algorithm;
FIG. 3 is a schematic diagram of a cascading failure critical triggering branch prediction system according to the present invention;
fig. 4 is a schematic structural diagram of an electronic device according to the present invention.
Detailed Description
The present invention will be described in detail below with reference to the embodiments with reference to the attached drawings. It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict.
The following detailed description is exemplary in nature and is intended to provide further details of the invention. Unless otherwise defined, all technical terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of exemplary embodiments according to the invention.
In order to enable the prediction of the weak branch to be applied in a real-time operation state, especially in a scene of topology change, a method capable of identifying the weak branch according to a topological structure and a power flow state is urgently required to be developed. The prediction process of the weak triggering branch circuit of the cascading failure can be regarded as a high-dimensional expression of the local operation state and the topological structure characteristics of the power grid, so that the topological structure and the electrical characteristics can be learned by using an algorithm of a neural network, and the weak branch circuit which can generate serious cascading failure can be predicted.
As shown in fig. 1, a first object of the present invention is to provide a method for predicting a critical triggering branch of a cascading failure, which includes the following steps:
acquiring online prediction target information;
constructing online prediction target information into a graph data sample set, and integrating the trend and the topological state into a branch characteristic matrix and an incidence matrix;
and inputting the branch characteristic matrix and the incidence matrix into a cascading failure simulation model to identify weak branches which possibly cause serious cascading failures, and obtaining the weak degree of each branch in the power system to be predicted.
According to the method, aiming at a topological structure and an operation state of a certain power grid, weak lines which may trigger serious cascading failures are predicted, and weak cause of the lines is analyzed. The method is applied on line aiming at different topological structures, has good generalization performance and interpretability, and has great engineering application value and popularization prospect.
Preferably, the method further comprises the steps of performing interpretable analysis on the weakness degree of each branch in the power system to be predicted, and solving to obtain the weakness cause of the predicted branch.
The technical scheme of the invention is further described in detail by combining the attached figure 2:
the method specifically comprises the following steps:
an off-line training link:
s1, inputting reference training target information, wherein the specific content comprises a complete topological structure of a target network and basic power flow state information;
further, inputting reference training target information, wherein the specific content comprises parameters such as a complete topological structure of a target network, parameters of each element, injected active power, injected reactive power, active load, reactive load of each node, voltage and phase angle of each node and the like;
s2, constructing a cascading failure simulation model, and generating cascading failure accident chain data according to the reference training target information;
generating cascading failure accident chain data by modifying a power grid topological structure and a tidal current state and performing cascading failure simulation according to the reference training target information;
further, the building of the cascading failure simulation model in step S2 and the generating of the cascading failure accident chain data from the reference training target information specifically include:
s21, aiming at topology change conditions such as branch maintenance and the like, constructing a new topology structure by breaking one or two branches at random;
s22, under the topological structure constructed in the step S21, aiming at the situations of load change, generator output adjustment, reactive device switching and the like in an actual power grid, injecting active power and reactive power into each node, and performing random step fluctuation adjustment on the active load and the reactive load to generate and record an initial tidal current running state;
s23, under the initial power flow operation state of the step S22, selecting a certain branch as a cascading failure triggering branch, and disconnecting the branch;
s24, carrying out power flow calculation on the power system, and carrying out load shedding operation on the node when the voltage of the node is lower than a safety threshold value, wherein the load shedding amount is determined by a dichotomy;
s25, carrying out power system load flow calculation, calculating branch load flow calculation branch outage probability, and disconnecting a branch when the branch outage probability is found to be greater than a threshold value, wherein the branch load flow calculation branch outage probability is represented by the formula (1):
Figure BDA0003180862140000081
in the formula, Pl、PmaxK is the transmission power of the branch, the upper limit of the transmission power of the branch, the limit transmission power multiple, pwFor branch protection stealth failure probability, pcThe branch outage probability when the power reaches the limit power.
S26, repeating the step S24 and the step S25 until the branch outage probability is larger than the threshold value and the node voltage is lower than the safety threshold value, and recording fault chain data;
s27, traversing all the branches of the power grid as cascading failure triggering branches, repeating the steps from S23 to S26, and recording cascading failure accident chain data triggered by each branch of the whole power grid under the initial tide running state of the step S22;
s28, repeating the steps S1 to S27, and constructing and recording cascading failure accident chain data under various topological structures and power flow states.
S3, collating cascading failure accident chain data, taking the trend and topological state information as input, taking branch weak indexes of all branches in the state as output, and generating a training sample;
as a preferred embodiment, generating the training sample specifically includes:
constructing cascading failure accident chain data into a graph data sample set required by a training model, constructing branch weak indexes of a branch by adopting cascading failure chain loss triggered by the branch in a certain state, and combining trend and topological state to form a training sample
Figure BDA0003180862140000082
And
Figure BDA0003180862140000083
and the branch characteristic matrix, the incidence matrix and the branch weak index are expressed under the topological state that 0,1 and 2 branches are banned respectively.
Wherein the input of the training set is defined as a branch characteristic matrix XGCNIIAnd the incidence matrix AGCNII. Defining the training set output as the branch weakness YGCNII。
Branch feature matrix XGCNIICovering the tidal current state information, expressed as:
XGCNII＝[xrelay,xp,xratio,xl,xloss,xfv,xftheta,xfpgen,xtv,xttheta,xtpload] (2)
wherein xrelay、xp、xratioThe branch transmission power, the limit transmission power and the ratio of the two represent the influence of overload protection and hidden fault xl、xlossRespectively defining a source end node, a tail end node and x according to the branch power flow direction for branch electric sodium and active lossfv,xftheta,xfpgenInjecting active power for voltage amplitude, voltage phase angle and power generation of a branch source end node respectively; x is the number oftv,xttheta,xtploadAnd respectively outputting active power for the voltage amplitude, the voltage phase angle and the load of the tail end node of the branch circuit.
Incidence matrix AGCNIITopology information is covered, represented as an N0-1 matrix:
Figure BDA0003180862140000091
the k weak categories defined are shown in table 1.
Table 1 output class description
Figure BDA0003180862140000092
Output YGCNII∈RN×kDividing the weak degree of the N branches into k types from small to large:
Figure BDA0003180862140000093
s4, building a weak branch prediction model based on a depth map convolutional neural network GCNII, and training the model by using the image data sample set in the step S3;
preferably, a depth map convolutional neural network for weak line prediction is constructed and trained, wherein the depth map convolutional neural network model comprises six graph convolutional GCNII layers, a full connection layer and a softmax output layer, and PReLU is selected as an activation function. Training the model based on the sample set, calculating the deviation between the predicted value and the label value by using the negative log-likelihood, and respectively obtaining YGCNIISet weight value omega for each categoryi(i ═ 0,1,2,3,4,5) to avoid the effects of biased samples; model parameters were updated by back-propagation, using an Adam optimizer, hyper-parameter optimization resulting in a learning rate of 0.005 and a batch size (batch size) of 32.
Online predictive analysis:
s5, inputting online prediction target information after the training of the prediction model is finished, wherein the specific content comprises the real-time topological structure and the load flow state information of the network to be predicted, namely calculating the weak degree of each branch in the state through the model, and identifying weak branches which possibly cause serious cascading failure;
the method specifically comprises the following steps:
acquiring online prediction target information, wherein the specific content comprises a complete topological structure of a power system to be predicted, parameters of each element, injected active power, injected reactive power, active load and reactive load of each node, and voltage and phase angle of each node;
constructing prediction input data, and sorting online prediction target information into branch characteristic matrix XGCNIIAnd the incidence matrix AGCNII；
Inputting the prediction input data into a trained prediction model, namely obtaining the weakness degree of each branch in the power system to be predicted through the model;
and S6, performing interpretability analysis on the prediction model by adopting a GNNExplainer algorithm, and solving and calculating the weak cause of the identified branch.
Further, performing interpretability analysis on the predicted branch weakness result, and selecting a mapping sample (Y)GCNII,XGCNII,AGCNII) Using GNN Expiainer algorithm as a prediction result
Figure BDA0003180862140000101
Generated interpretation subset
Figure BDA0003180862140000102
Wherein
Figure BDA0003180862140000103
Respectively represent the pair predictions
Figure BDA0003180862140000104
Neighborhood sets and feature subsets with significant impact. The GNN Explainer optimizes the structure mask σ (M) and feature mask σ (F) of the entire computation graph, respectively, with the goal of mutual information maximization based on the mutual information model MI:
Figure BDA0003180862140000105
where Y is the distribution of the predicted values for a branch, H (g) represents the entropy calculation for a discrete quantity, M ∈ RN×NFor the computed sub-graph mask parameters to be learned, AcIs a adjacency matrix, C represents the number of classes of the classification problem, σ (g) is a sigmond function, XS＝{xjvj∈GSIndicates neighborhood branch set GSAll branch characteristics. The above equation is limited to quantizing the computational domain information with MI
Figure BDA0003180862140000106
Time, predicted value
Figure BDA0003180862140000107
Based on the method of counterintuitive interpretation, consider the missing prediction values of other branches and their partial features
Figure BDA0003180862140000108
Extracting and predicting values of
Figure BDA0003180862140000111
Graph domain interpretation subset with highest mutual information
Figure BDA0003180862140000112
Thus, a mapping sample (Y) is selectedGCNII,XGCNII,AGCNII) Optimizing according to the formula (5), namely obtaining Y mutual information with the predicted value of the current model and obtaining the weak cause of the predicted branch
Figure BDA0003180862140000113
As shown in fig. 3, another objective of the present invention is to provide a cascading failure critical triggering branch prediction system, which includes:
the information acquisition module is used for acquiring online prediction target information comprising a complete topological structure of a target network and basic power flow state information;
the data construction module is used for arranging the online prediction target information into a branch characteristic matrix and an incidence matrix so as to construct prediction input data;
and the branch prediction module is used for identifying weak branches which possibly cause serious cascading failure from the prediction input data through the cascading failure simulation model to obtain the weak degree of each branch in the power system to be predicted.
Preferably, the system further comprises a cause analysis module, configured to perform interpretable analysis on the weakness degree of each branch in the power system to be predicted, and solve to obtain the weakness cause of the predicted branch.
A third object of the present invention is to provide an electronic device, as shown in fig. 4, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the cascading failure critical triggering branch prediction method when executing the computer program.
The cascading failure key triggering branch prediction method comprises the following steps:
acquiring online prediction target information;
constructing online prediction target information into a graph data sample set, and integrating the trend and the topological state into a branch characteristic matrix and an incidence matrix;
and inputting the branch characteristic matrix and the incidence matrix into a cascading failure simulation model to identify weak branches which possibly cause serious cascading failures, and obtaining the weak degree of each branch in the power system to be predicted.
A fourth object of the present invention is to provide a computer-readable storage medium, which stores a computer program, which when executed by a processor, implements the steps of the cascading failure critical triggering branch prediction method.
The cascading failure key triggering branch prediction method comprises the following steps:
acquiring online prediction target information;
constructing online prediction target information into a graph data sample set, and integrating the trend and the topological state into a branch characteristic matrix and an incidence matrix;
and inputting the branch characteristic matrix and the incidence matrix into a cascading failure simulation model to identify weak branches which possibly cause serious cascading failures, and obtaining the weak degree of each branch in the power system to be predicted.
As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solutions of the present invention and not for limiting the same, and although the present invention is described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: modifications and equivalents may be made to the embodiments of the invention without departing from the spirit and scope of the invention, which is to be covered by the claims.

Claims (13)

1. A prediction method for a critical trigger branch of cascading failures is characterized by comprising the following steps:
acquiring online prediction target information;
constructing online prediction target information into a graph data sample set, and integrating the trend and the topological state into a branch characteristic matrix and an incidence matrix;
and inputting the branch characteristic matrix and the incidence matrix into a cascading failure simulation model to identify weak branches which possibly cause serious cascading failures, and obtaining the weak degree of each branch in the power system to be predicted.
2. The method of claim 1, wherein:
the online prediction target information comprises one or more of a complete topological structure of the power system to be predicted, parameters of each element, injected active power, injected reactive power, active load, reactive load of each node, and voltage and phase angle of each node.
3. The method of claim 1, wherein:
the building process of the cascading failure simulation model comprises the following steps:
acquiring reference training target information, wherein the target information comprises a complete topological structure of a target network and basic power flow state information;
generating cascading failure accident chain data by the reference training target information; arranging cascading failure accident chain data, taking the information of the trend and the topological state as input, taking the branch weak index of each branch in the state as output, and generating a training sample set;
and constructing and training a cascading failure simulation model based on the depth map convolutional neural network.
4. The method of claim 3, wherein:
the cascading failure accident chain data generated by the reference training target information is generated by modifying a power grid topological structure and a tide state and performing cascading failure simulation.
5. The method of claim 4, wherein:
the generating of the cascading failure accident chain data specifically includes:
aiming at the branch topology change condition, constructing a new topology structure by randomly breaking one or two branches;
under the constructed topological structure, aiming at the situations of load change, generator output regulation and reactive device switching in an actual power grid, injecting active power and reactive power into each node, and carrying out random graded fluctuation regulation on the active load and the reactive load to generate and record an initial tide running state;
under the initial tide running state, selecting a certain branch as a cascading failure triggering branch, and disconnecting the branch;
carrying out load shedding operation on the node when the load shedding operation is carried out on the node when the node voltage is lower than the safety threshold value, wherein the load shedding amount is determined by a dichotomy; calculating branch load flow to calculate the branch outage probability, and disconnecting the branch when the branch outage probability is found to be greater than a threshold value; recording fault chain data until no branch outage probability is greater than a threshold value and the node voltage is lower than a safety threshold value;
traversing all branches of the power grid as cascading failure triggering branches to obtain cascading failure chain data triggered by each branch of the whole power grid in an initial tide running state;
and then cascading failure chain data under various topological structures and tidal current states are constructed and recorded.
6. The method of claim 5, wherein:
the branch outage probability in the power system load flow calculation is obtained by the following method:
Figure FDA0003180862130000021
in the formula, Pl、PmaxK is the transmission power of the branch, the upper limit of the transmission power of the branch, the limit transmission power multiple, pwFor branch protection stealth failure probability, pcThe branch outage probability when the power reaches the limit power.
7. The method of claim 3, wherein:
the sorting of cascading failure accident chain data specifically comprises:
the cascading failure accident chain data are constructed into a graph data sample set, the branch weak index of a branch is constructed by adopting the cascading failure chain loss triggered by the branch in a certain state, the load flow and the topological state are integrated to form a training sample, and the training sample comprises a branch characteristic matrix, an incidence matrix and a branch weak index.
8. The method of claim 7, wherein:
branch feature matrix XGCNIICovering the tidal current state information, expressed as:
XGCNII＝[xrelay，xp，xratio，xl，xloss，xfv，xftheta，xfpgen，xtv，xttheta，xtpload]
wherein x isrelay、xp、xratioThe branch transmission power, the limit transmission power and the ratio of the two represent the influence of overload protection and hidden fault xl、xlossRespectively defining a source end node, a tail end node and x according to the branch power flow direction for branch electric sodium and active lossfv，xftheta，xfpgenInjecting active power for voltage amplitude, voltage phase angle and power generation of a branch source end node respectively; x is the number oftv，xttheta，xtploadRespectively outputting active power for the voltage amplitude, the voltage phase angle and the load of the tail end node of the branch circuit;
incidence matrix AGCNIICoveringTopology information, expressed as a 0-1 matrix of NxN:
Figure FDA0003180862130000031
branch weakness index YGCNII∈RN×kDividing the weak degree of the N branches into k types from small to large:
Figure FDA0003180862130000032
9. the method of claim 1, wherein:
performing interpretable analysis on the weakness degree of each branch in the power system to be predicted, and solving to obtain the weakness cause of the predicted branch; the method specifically comprises the following steps:
selecting a mapping sample (Y)GCNII，XGCNII，AGCNII) Using GNN Expiainer algorithm as prediction result
Figure FDA0003180862130000036
Generating a subset of interpretations
Figure FDA0003180862130000033
Wherein
Figure FDA0003180862130000034
Respectively represent the pair predictions
Figure FDA0003180862130000037
Neighborhood branch sets and feature subsets with significant impact;
the GNN Explainer algorithm optimizes the structure mask σ (M) and feature mask σ (F) of the entire computation graph, respectively, with the goal of mutual information maximization based on the mutual information model MI:
Figure FDA0003180862130000035
where Y is the distribution of the predicted values for a branch, H (g) represents the entropy calculation for a discrete quantity, M ∈ RN×NFor the computed sub-graph mask parameters to be learned, AcIs a adjacency matrix, C represents the number of classes of the classification problem, σ (g) is a sigmond function, Xs＝{xjvj∈GsIndicates neighborhood branch set GsAll branch characteristics;
optimizing to obtain mutual information with the predicted value of the current model, extracting and predicting the value
Figure FDA0003180862130000041
Graph domain interpretation subset with highest mutual information
Figure FDA0003180862130000042
Thereby obtaining the weak cause of the predicted branch
Figure FDA0003180862130000043
10. A cascading failure critical triggering branch prediction system, comprising:
the information acquisition module is used for acquiring online prediction target information comprising a complete topological structure of a target network and basic power flow state information;
the data construction module is used for arranging the online prediction target information into a branch characteristic matrix and an incidence matrix so as to construct prediction input data;
and the branch prediction module is used for identifying weak branches which possibly cause serious cascading failure from the prediction input data through the cascading failure simulation model to obtain the weak degree of each branch in the power system to be predicted.
11. The cascading failure critical triggering branch prediction system of claim 10, further comprising:
and the cause analysis module is used for performing interpretable analysis on the weakness degree of each branch in the power system to be predicted and solving to obtain the weakness cause of the predicted branch.
12. An electronic device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing the steps of the cascading failure critical triggering branch prediction method of any one of claims 1-9 when executing the computer program.
13. A computer readable storage medium storing a computer program which, when executed by a processor, implements the steps of the cascading failure critical triggering branch prediction method of any one of claims 1-9.
CN202110846723.3A 2021-07-26 2021-07-26 Method, system, equipment and storage medium for predicting critical trigger branch of cascading failure Active CN113505458B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110846723.3A CN113505458B (en) 2021-07-26 2021-07-26 Method, system, equipment and storage medium for predicting critical trigger branch of cascading failure

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110846723.3A CN113505458B (en) 2021-07-26 2021-07-26 Method, system, equipment and storage medium for predicting critical trigger branch of cascading failure

Publications (2)

Publication Number Publication Date
CN113505458A true CN113505458A (en) 2021-10-15
CN113505458B CN113505458B (en) 2024-07-02

Family

ID=78014621

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110846723.3A Active CN113505458B (en) 2021-07-26 2021-07-26 Method, system, equipment and storage medium for predicting critical trigger branch of cascading failure

Country Status (1)

Country Link
CN (1) CN113505458B (en)

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113987852A (en) * 2021-12-29 2022-01-28 长沙理工大学 High-risk circuit combination analysis method for electric power information physical system
CN114154558A (en) * 2021-11-12 2022-03-08 山东浪潮科学研究院有限公司 Distributed energy power generation load prediction system and method based on graph neural network
CN114548762A (en) * 2022-02-22 2022-05-27 浙江大学 Real-time power system cascading failure risk assessment method and system based on space-time diagram neural network
CN115865727A (en) * 2022-11-24 2023-03-28 西南交通大学 Branch correlation risk assessment method based on credibility and credibility inference graph
CN116401614A (en) * 2023-06-06 2023-07-07 苏州振州机电科技有限公司 Equipment fault identification method and system
CN116885717A (en) * 2023-09-08 2023-10-13 北京龙德缘电力科技发展有限公司 Multistage topology analysis method for power system
CN117054805A (en) * 2023-07-10 2023-11-14 国网湖北省电力有限公司超高压公司 DGCN network-based fault diagnosis method for gas-insulated switchgear
CN118245832A (en) * 2024-05-28 2024-06-25 国网山东省电力公司阳信县供电公司 Fault power failure information generation method, system, electronic equipment and storage medium

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106998064A (en) * 2017-04-10 2017-08-01 清华大学深圳研究生院 A kind of cascading failure fault chains searching method
CN110675043A (en) * 2019-09-17 2020-01-10 深圳供电局有限公司 Method and system for determining power grid power failure key line based on cascading failure model
WO2020078109A1 (en) * 2018-10-17 2020-04-23 中国电力科学研究院有限公司 Method, device, and storage medium for identifying weak section of electrical power grid

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106998064A (en) * 2017-04-10 2017-08-01 清华大学深圳研究生院 A kind of cascading failure fault chains searching method
WO2020078109A1 (en) * 2018-10-17 2020-04-23 中国电力科学研究院有限公司 Method, device, and storage medium for identifying weak section of electrical power grid
CN110675043A (en) * 2019-09-17 2020-01-10 深圳供电局有限公司 Method and system for determining power grid power failure key line based on cascading failure model

Cited By (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114154558A (en) * 2021-11-12 2022-03-08 山东浪潮科学研究院有限公司 Distributed energy power generation load prediction system and method based on graph neural network
CN114154558B (en) * 2021-11-12 2024-05-21 山东浪潮科学研究院有限公司 Distributed energy power generation load prediction system and method based on graph neural network
CN113987852A (en) * 2021-12-29 2022-01-28 长沙理工大学 High-risk circuit combination analysis method for electric power information physical system
CN114548762A (en) * 2022-02-22 2022-05-27 浙江大学 Real-time power system cascading failure risk assessment method and system based on space-time diagram neural network
CN115865727B (en) * 2022-11-24 2024-04-12 西南交通大学 Branch association risk assessment method based on credibility and non-credibility inference graph
CN115865727A (en) * 2022-11-24 2023-03-28 西南交通大学 Branch correlation risk assessment method based on credibility and credibility inference graph
CN116401614B (en) * 2023-06-06 2023-08-18 苏州振州机电科技有限公司 Equipment fault identification method and system
CN116401614A (en) * 2023-06-06 2023-07-07 苏州振州机电科技有限公司 Equipment fault identification method and system
CN117054805A (en) * 2023-07-10 2023-11-14 国网湖北省电力有限公司超高压公司 DGCN network-based fault diagnosis method for gas-insulated switchgear
CN117054805B (en) * 2023-07-10 2024-10-29 国网湖北省电力有限公司超高压公司 Gas-insulated switchgear fault diagnosis method based on DGCN network
CN116885717A (en) * 2023-09-08 2023-10-13 北京龙德缘电力科技发展有限公司 Multistage topology analysis method for power system
CN116885717B (en) * 2023-09-08 2023-11-21 北京龙德缘电力科技发展有限公司 Multistage topology analysis method for power system
CN118245832A (en) * 2024-05-28 2024-06-25 国网山东省电力公司阳信县供电公司 Fault power failure information generation method, system, electronic equipment and storage medium

Also Published As

Publication number Publication date
CN113505458B (en) 2024-07-02

Similar Documents

Publication Publication Date Title
CN113505458B (en) Method, system, equipment and storage medium for predicting critical trigger branch of cascading failure
Shi et al. Artificial intelligence techniques for stability analysis and control in smart grids: Methodologies, applications, challenges and future directions
Pourdaryaei et al. Short-term electricity price forecasting via hybrid backtracking search algorithm and ANFIS approach
Ghods et al. Methods for long-term electric load demand forecasting; a comprehensive investigation
Liu et al. Searching for critical power system cascading failures with graph convolutional network
Kaboli et al. An expression-driven approach for long-term electric power consumption forecasting
CN116245033A (en) Artificial intelligent driven power system analysis method and intelligent software platform
Deng et al. Short-term load forecasting by using improved GEP and abnormal load recognition
CN104732067A (en) Industrial process modeling forecasting method oriented at flow object
Zhang et al. Application and progress of artificial intelligence technology in the field of distribution network voltage Control: A review
Cui et al. A frequency domain approach to predict power system transients
Li et al. A probabilistic data-driven method for response-based load shedding against fault-induced delayed voltage recovery in power systems
Liu et al. Analysis and prediction of power distribution network loss based on machine learning
CN116826743A (en) Power load prediction method based on federal graph neural network
CN116826733A (en) Photovoltaic power prediction method and system
CN116780509A (en) Power grid random scene generation method integrating discrete probability and CGAN
Wu et al. Fast dc optimal power flow based on deep convolutional neural network
Hayes et al. Multi-objective coordination graphs for the expected scalarised returns with generative flow models
JP2023010660A (en) Method for predicting c axial length of crystal structure of lithium compound, method for building learning model, and system for predicting crystal structure having maximum c axial length
Tao et al. Optimization of green agri-food supply chain network using chaotic PSO algorithm
CN113283638A (en) Load extreme curve prediction method and system based on fusion model
CN111950765A (en) Probabilistic transient stability prediction method based on stacked noise reduction self-encoder
Xu et al. Coordinated preventive-corrective control for power system transient stability enhancement based on machine learning-assisted optimization
CN118899844A (en) Power distribution network load transfer control method and system based on neural network decision distillation
Muller Creating building energy prediction models with convolutional recurrent neural networks

Legal Events

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