CN112990551A - Cascading failure evolution path tracing and predicting method and device based on knowledge graph - Google Patents
Cascading failure evolution path tracing and predicting method and device based on knowledge graph Download PDFInfo
- Publication number
- CN112990551A CN112990551A CN202110191955.XA CN202110191955A CN112990551A CN 112990551 A CN112990551 A CN 112990551A CN 202110191955 A CN202110191955 A CN 202110191955A CN 112990551 A CN112990551 A CN 112990551A
- Authority
- CN
- China
- Prior art keywords
- event
- characteristic
- events
- cascading failure
- evolution
- 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.)
- Withdrawn
Links
- 238000000034 method Methods 0.000 title claims abstract description 42
- 230000007246 mechanism Effects 0.000 claims abstract description 7
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 6
- 238000010219 correlation analysis Methods 0.000 claims abstract description 6
- 238000012549 training Methods 0.000 claims description 21
- 238000004458 analytical method Methods 0.000 claims description 10
- 230000008569 process Effects 0.000 claims description 8
- 238000010845 search algorithm Methods 0.000 claims description 6
- 239000000203 mixture Substances 0.000 claims description 5
- 230000008859 change Effects 0.000 claims description 4
- 238000012706 support-vector machine Methods 0.000 claims description 4
- 230000001052 transient effect Effects 0.000 claims description 4
- 238000012163 sequencing technique Methods 0.000 claims description 2
- 238000010801 machine learning Methods 0.000 abstract description 5
- 239000000284 extract Substances 0.000 abstract description 2
- 238000012502 risk assessment Methods 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 8
- 238000004590 computer program Methods 0.000 description 7
- 230000000903 blocking effect Effects 0.000 description 6
- 230000006870 function Effects 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000011217 control strategy Methods 0.000 description 2
- 230000005684 electric field Effects 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000001364 causal effect Effects 0.000 description 1
- 230000019771 cognition Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000003211 malignant effect Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 239000009360 ningdong Substances 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000010355 oscillation Effects 0.000 description 1
- 238000012358 sourcing Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/02—Knowledge representation; Symbolic representation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/04—Inference or reasoning models
- G06N5/041—Abduction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- General Engineering & Computer Science (AREA)
- Marketing (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- General Business, Economics & Management (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Tourism & Hospitality (AREA)
- Mathematical Physics (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a cascading failure evolution path tracing and predicting method and device based on a knowledge graph, which can be applied to an alternating current-direct current power grid cascading failure risk assessment and early warning system. The method extracts key characteristic quantities by combining a physical mechanism and a correlation analysis method, automatically learns and judges the incidence relation between characteristic events through a machine learning algorithm, realizes the matching and judgment of the upper and lower incidence relation of the characteristic events by constructing a knowledge graph of two levels, judges the result and the time sequence characteristics according to the incidence relation of the characteristic events, and identifies the cascading failure evolution path based on a depth-first search strategy. The method introduces an intelligent learning and analyzing engine to replace thinking and judgment of scheduling personnel, and provides effective technical support for tracing and prejudging the complex cascading failure.
Description
Technical Field
The invention belongs to the technical field of power system automation, and particularly relates to a cascading failure evolution path tracing and predicting method based on a knowledge graph, and further relates to a cascading failure evolution path tracing and predicting device based on the knowledge graph.
Background
With the large amount of new energy such as wind power, photovoltaic and the like being connected to the power grid, the direct-current cross-district power transmission scale continuously increases, the chain reaction evolution of local faults of the power grid becomes the global safety risk characteristic which is increasingly obvious, in typical cases, multiple direct-current commutation failures are caused by alternating-current short circuit faults of a cross-district receiving-end power grid, the large-scale fluctuation of the power and the voltage of a sending-end alternating-current connecting line is caused by high-power impact during the commutation failures, and further serious events threatening the safe and stable operation of the global power grid, such as power grid disconnection, new energy large-scale off-grid. Based on the power grid characteristic events, the chain accident tracing analysis and the advanced prediction are carried out, and the method has important significance for blocking the malignant evolution of the chain accidents and accident disposal.
The cascading failure evolution path is closely related to a power grid operation mode, a grid-related protection of equipment and other factors, and strong uncertainty exists. The cognition of the scheduling operator on the cascading failure evolution process mainly comes from an off-line mode calculation analysis conclusion and event alarm information in the scheduling automation system. However, all operation scenes are difficult to be exhausted by off-line mode calculation, and off-line knowledge is used for tracing characteristic events, so that the accuracy of a conclusion is difficult to guarantee. The existing fault diagnosis or comprehensive intelligent alarm system can realize the identification of single equipment fault through multi-source data integration and expert experience, and cannot judge the linkage incidence relation among a plurality of events. The occurrence of cascading failures of a power system and the evolution of events have causal relationship and timing characteristics in nature, and key characteristic events have traceability and predictability, which are extremely dependent on operation experience and knowledge accumulation. The knowledge graph is a data structure based on a graph, is used for describing the relation among knowledge, has the capabilities of retrieving, reasoning and analyzing the knowledge, and is particularly suitable for being used as a reasoning analysis engine of rule knowledge.
Disclosure of Invention
The invention aims to overcome the defects in the prior art, and provides a cascading failure evolution path tracing and predicting method based on a knowledge graph.
In order to solve the technical problems, the technical scheme of the invention is as follows.
In a first aspect, the invention provides a cascading failure evolution path tracing and predicting method based on a knowledge graph, which comprises the following processes:
acquiring cascading failure historical data and generating a cascading failure evolution sample database;
training to obtain a characteristic event relation model based on the obtained cascading failure evolution sample database, wherein the input of the characteristic event relation model is the key characteristic quantity of an event pair, and the output of the characteristic event relation model is whether the relation between the event pair is established or not;
extracting event classes, characteristic events, relations among the characteristic events and relation attributes from a cascading failure evolution sample database to form a characteristic event evolution knowledge graph, wherein the relations among the characteristic events are obtained by inputting any characteristic event composition event into a trained characteristic event relation model;
and searching the characteristic event evolution knowledge graph aiming at the obtained actually measured characteristic event sequence to obtain the tracing and prediction path of the cascading failure characteristic event.
Optionally, the cascading failure history data includes an event sequence arranged according to an event occurrence time sequence, power grid operation mode data before and after an event, power grid transient electrical quantity change information caused by the event, a failure location, and a failure type.
Optionally, the training process of the feature event relationship model includes:
determining a training event pair set from a cascading failure evolution sample database through physical mechanism analysis;
extracting key characteristic quantity of each event pair set by a data correlation analysis method;
and training based on a support vector machine algorithm by taking the key characteristic quantity as input and whether the relation between the event pairs is established as output to obtain a characteristic event relation model.
Optionally, the extracting the event class, the feature event, the relationship between the feature events, and the key feature quantity to form a feature event evolution knowledge graph includes:
taking the event class as a node in the top layer;
taking the specific characteristic event as a bottom layer node, and connecting the specific characteristic event with the event class belonging to the upper layer node by using a connecting line;
in the bottom node, if the events have a relationship, connecting the two events in the graph by using an arrow connecting line, and pointing the arrow to a subsequent event;
and taking the relation attribute of the event as a node led out by the bottom node.
Optionally, the relationship between the characteristic events further comprises a regularization knowledge derivation.
Optionally, the searching a feature event evolution knowledge graph for the obtained actually measured feature event sequence to obtain a tracing path and a prediction path of the cascading failure feature event includes:
arranging the actually measured characteristic event sequence according to a time sequence to form a characteristic event set;
removing feature events which are not matched with the feature event evolution knowledge graph from the feature event set;
searching all related characteristic events upwards by adopting a depth-first search algorithm and taking the sequenced last characteristic event in the characteristic event set as a starting point;
forming event pairs by all searched characteristic events according to a search sequence, inputting the event pairs into a characteristic event relation model to obtain the relation between the characteristic events, and if the relation between the characteristic events is 'true' and the events exist in a characteristic event set, bringing the event pairs into a tracing path set;
sequencing the characteristic events in the tracing path set according to a time sequence, wherein the sequenced event sequence is the tracing path of the cascading failure;
a depth-first search algorithm is adopted, the last sequenced characteristic event in the characteristic event set is taken as a starting point, and all the related characteristic events are searched downwards;
forming event pairs by all searched characteristic events according to a search sequence, inputting the event pairs into a characteristic event relation model to obtain the relation between the characteristic events, and if the relation between the characteristic events is 'true' and the events exist in a characteristic event set, bringing the event pairs into a prediction path set;
and the event sequence in the predicted path set is the predicted cascading failure evolution path.
Optionally, the removing, from the feature event set, feature events that do not have matching in the feature event evolution knowledge graph includes:
preferentially searching whether a matched event class exists in the feature event evolution knowledge graph or not aiming at each feature event in the feature event set, and if not, removing the feature event from the feature event set;
and if the event class to which the characteristic event belongs exists, matching the key characteristic quantity of the event characteristic with the relationship attributes of the event characteristics and the event characteristics in the event class aiming at the event characteristic, and if the matching is unsuccessful, removing the characteristic event from the characteristic event set.
In a second aspect, the invention provides a cascading failure evolution path tracing and predicting device based on a knowledge graph, which comprises:
the data acquisition module is used for acquiring cascading failure historical data and generating a cascading failure evolution sample database;
the model training module is used for training to obtain a characteristic event relation model based on the obtained cascading failure evolution sample database, wherein the input of the characteristic event relation model is the key characteristic quantity of an event pair, and the output of the characteristic event relation model is whether the relation between the event pairs is established or not;
the knowledge graph building module is used for extracting event types, characteristic events, relations among the characteristic events and relation attributes from a cascading failure evolution sample database to form a characteristic event evolution knowledge graph, wherein the relations among the characteristic events are obtained by inputting a trained characteristic event relation model by any characteristic event composition event;
and the tracing prediction module is used for searching the characteristic event evolution knowledge graph aiming at the obtained actually measured characteristic event sequence to obtain the tracing and prediction path of the cascading failure characteristic event.
Compared with the prior art, the invention has the following beneficial effects: according to the invention, the key characteristic quantity is extracted from the power grid operation state quantity and the fault information, and the correlation relation of the characteristic events is automatically learned and judged through machine learning, so that the accuracy of event relation identification is effectively improved. And matching and judging the upper and lower level incidence relation of the characteristic event by constructing a two-level knowledge map. And identifying a cascading failure evolution path based on a depth-first search strategy according to the characteristic event incidence relation judgment result and the time sequence characteristic, and meeting the requirements of online tracing analysis and advanced prediction of a cascading failure chain. The method utilizes knowledge graph technology to replace thinking and judgment of scheduling personnel, and provides effective technical support for tracing and prejudging the difficult problems of complex cascading failures.
Drawings
FIG. 1 is a flow chart showing the method of the present invention;
FIG. 2 is a knowledge-graph embodiment of the present invention;
FIG. 3 is an example of the display of measured occurrences in a knowledge graph in accordance with the present invention.
Detailed Description
The invention is further described below with reference to the accompanying drawings. The following examples are only for illustrating the technical solutions of the present invention more clearly, and the protection scope of the present invention is not limited thereby.
The invention provides a cascading failure evolution path tracing and predicting method based on a knowledge graph, which can be applied to an alternating current-direct current power grid cascading failure risk assessment and early warning system. The method comprises the steps of taking power grid running state quantity and fault information as input characteristic quantities, extracting key characteristic quantities by combining a physical mechanism and a correlation analysis method, automatically learning and judging incidence relations among characteristic events through a machine learning algorithm, realizing matching and judgment of upper and lower incidence relations of the characteristic events by constructing a knowledge graph of two levels, judging results and time sequence characteristics according to the incidence relations of the characteristic events, and identifying a cascading fault evolution path based on a depth-first search strategy. The method introduces an intelligent learning and analyzing engine to replace thinking and judgment of scheduling personnel, and provides effective technical support for tracing and prejudging the complex cascading failure.
The invention discloses a cascading failure evolution path tracing and predicting method based on a knowledge graph, which is shown in a figure 1 and comprises the following steps:
step S1: generating a cascading failure evolution sample database based on the cascading failure historical data, and entering step S2;
the cascading failure historical data comprises an event sequence arranged according to an event occurrence sequence, power grid operation mode data before and after an event, power grid transient electric quantity change information caused by the event, a failure site and a failure type.
Step S2: training a characteristic event relation model based on a support vector machine algorithm based on the cascading failure evolution sample database constructed in the step S1, obtaining a characteristic event, a characteristic event relation model, a key characteristic quantity and regularization knowledge from the sample database constructed in the step S1 through knowledge extraction and knowledge representation, storing the characteristic event, the characteristic event relation model, the key characteristic quantity and the regularization knowledge into a knowledge database, forming a characteristic event evolution knowledge map through designing a cascading failure evolution knowledge map body, and entering the step S3;
the feature event relation model refers to a mathematical model obtained through machine learning training to describe the relation between input and output, and the input of the model is: the method comprises the following steps of (1) arranging an event sequence according to an event occurrence time sequence, power grid operation mode data, power grid transient electric quantity change information caused by events, fault sites and fault types; the output of the model is: the event relation is established or not established; the event relation specifically refers to whether the power grid causes a subsequent event to occur if the preceding event occurs under a certain condition. The establishment of an event relationship means that a subsequent event occurs if a preceding event occurs.
The specific steps of training the characteristic event relation model are as follows:
s2-1) determining an event pair set which needs to be trained by a characteristic event relation model through physical mechanism analysis, wherein the event pair set is the names of two types of events;
physical mechanism analysis is to analyze and find whether a relationship exists between different events through inherent characteristics of a power system, for example, event pairs are integrated into an alternating current short-circuit event and a direct current commutation failure event, the alternating current short-circuit event causes the voltage of a commutation bus at a direct current inversion side to be reduced, and a conduction valve cannot recover blocking capability under the action of reverse voltage, so that the direct current commutation failure event is caused, and therefore, a relationship exists between the alternating current short-circuit event and the direct current commutation failure event.
S2-2) extracting key characteristic quantities of each event pair set by a data correlation analysis method, wherein the characteristic quantities comprise power grid operation state quantities and fault information;
s2-3) taking the key characteristic quantity as input, taking whether the relation is established as output, and carrying out event correlation training based on a support vector machine algorithm to obtain a characteristic event relation model.
The knowledge extraction is oriented to text data, a pre-labeled corpus training model is utilized, the model learns the probability that commutation fails and a word off-line is used as an entity component, then a probability value of a candidate field serving as an entity is calculated, and events, relations and relation attributes are further extracted.
The knowledge representation takes an ontology as a core, takes a Resource Description Framework (RDF) as a basic framework, and describes events, relations and relation attributes through a network ontology language (OWL).
Characteristic events refer to events with causality, globality, and chronology, including: generator tripping, line tripping, transformer tripping, direct current commutation failure, direct current blocking, new energy off-line, safety control or system protection generator tripping, high cycle generator tripping, low frequency load shedding, splitting and power oscillation;
the key characteristic quantity refers to the power grid operation state quantity and event information which influence whether the characteristic event relation model is established or not;
regularization knowledge refers to equipment or device control strategies inherent to the power system and not affected by changes in the grid operating mode, including: and 3 times of failure of direct current continuous commutation triggers direct current blocking.
The knowledge graph ontology specifically comprises:
event classes, i.e., a set of similar characteristic events, such as commutation failure, wind farm offline, etc., are top-level nodes in the knowledge graph;
events, namely specific characteristic events, such as a Binjin direct-current commutation failure, a Linshao direct-current commutation failure, a Ningdong wind electric field off-line and the like, are bottom-layer nodes in the knowledge graph;
relationships, namely relationships among events, including true or false, are event relationships obtained based on a characteristic event relationship model, and are relationships among nodes in a knowledge graph and represented by arrows connected among the nodes;
and the relation attribute, namely the key characteristic quantity, is a node led out for the bottom-layer node in the knowledge graph.
Rules, that is, the inherent regularization knowledge of the power system, such as a safety control strategy, a dc control protection strategy, and the like, may also be expressed by the above-mentioned knowledge map ontology, and it is explained with reference to fig. 1 that the three times of dc commutation failure in the agile presentation result in the agile presentation of dc bipolar latch, that is, the regularization knowledge, the events are the agile presentation of dc commutation failure and the agile presentation of dc bipolar latch, the relationship is established or caused, and the relationship attributes are the three times of commutation failure. Rules are knowledge that does not require learning and training and are inherent logic. The two event relations of the regularization knowledge are certain established or not established, the judgment is easy, and the relation model training is not needed. But may also constitute a traceback and prediction path and therefore need to be listed. The path search process is equally applicable to regularized knowledge.
When the knowledge graph body is constructed, event classes are used as nodes in a top layer, such as commutation failure, direct current blocking, wind power plant off-grid and the like, specific events are used as nodes in a bottom layer and are connected with the event classes in the nodes in the upper layer through connecting lines, the specific events include Linshao direct current commutation failure, Linshao direct current blocking, Ningdongfeng electric field off-grid and the like, and the event classes and the events form a two-layer knowledge graph. In the underlying nodes, if there is a relationship between events (including true and false), two events (nodes) are connected in the graph with an arrow connecting line, and the arrow points to the subsequent event. The relationship attribute of the event is the node led out by the bottom node. An example of a knowledge graph is shown in fig. 2, where the knowledge graph is a node and a relationship between nodes, the relationship is represented by arrows connecting nodes in the graph, and the relationship attribute is also represented by a node.
When the specific events are detected by the power grid, the searching and matching of the event classes can be carried out in a priority mode, so that the searching range of the characteristic events and the relation model can be greatly reduced.
Step S3: acquiring power grid online operation mode data and actual measurement characteristic events pushed by the comprehensive intelligent alarm system, and entering step S4;
step S4: based on a depth-first search technology, a knowledge graph is used for carrying out online search on feature event evolution paths, and tracing and predicting paths of cascading failure feature events and related alarm information are output.
The specific steps of the online search of the characteristic event tracing evolution path are as follows:
s4-1) arranging a sequence of power grid actual measurement characteristic events according to time sequence to form a characteristic event set { Fi}；
S4-2) setting the tracing path set S 'and the evolution path set S' as empty sets;
s4-3) for each characteristic event in the characteristic event set, preferentially searching the event class in the knowledge graph, and if the event class to which the characteristic event belongs is not found in the knowledge graph, selecting the characteristic event from the set { F {iRemoving the characteristic events if the events belonging to the characteristic events are foundMatching key characteristic quantities such as power grid online operation mode data related to the event characteristics with relation attributes of each event and the event characteristics in the event class aiming at the event characteristics; if the matching is successful, retaining, if the matching is unsuccessful, selecting the characteristic event from the set { F }iRemoving;
matching means that the event names are consistent and the relationship attribute contents are consistent, otherwise, the event names are not matched.
S4-4) adopting a depth-first search algorithm to set FiSearching all the related characteristic events upwards (searching the preamble events) by taking the last sequenced characteristic event as a starting point;
s4-5) forming event pairs by all the searched characteristic events according to the search sequence, inputting the event pairs into a characteristic event relation model for judgment, and if the characteristic event relation model is output to be 'true' and the events exist in a set { F }iIn indicates that a preceding event results in a subsequent event, the event pair is included in S'; if the output of the characteristic event relation model is 'false', the fact that the preceding event does not cause the subsequent event is indicated;
s4-6) sorting the characteristic events in the S 'according to time sequence, wherein the sorted S' is a tracing path of the cascading failure;
s4-7) adopting a depth-first search algorithm to { F }iSearching all the related characteristic events downwards (searching subsequent events) by taking the last sequenced characteristic event as a starting point;
s4-8) forming event pairs by all the searched characteristic events according to the search sequence, inputting the event pairs into a characteristic event relation model for judgment, and if the characteristic event relation model is output to be 'true' and the events exist in a set { FiIn indicates that the preceding event results in a subsequent event, bringing the event pair into S "; if the output of the characteristic event relation model is 'false', the fact that the preceding event does not cause the subsequent event is indicated;
s4-9) the event sequence in S' is the predicted cascading failure evolution path.
Referring to fig. 3, it is assumed that the actually measured grid characteristic events are a, i, g, d, and e according to the occurrence time sequence, source tracing analysis is performed with e as a starting point, d and h are possible higher-level events, and according to the grid operating state and the actually measured event information, the relationship that "d causes e" is established, the higher-level event that d is e can be determined, and similarly, the higher-level event that g is d can be inferred, i is the higher-level event that g is g, and a does not cause b to occur, so that the evolution path is inferred to be (i, g, d, e), and the sourcing fault is i.
And (3) carrying out evolution prediction by taking e as a starting point, and if the relation of 'e causing f' is established and if f occurs, the relation of 'f causing p' is established, then the predicted cascading failure evolution path is (e, f, p).
The invention has the following beneficial effects:
the method extracts key characteristic quantities from the power grid operation state quantity and fault information by combining a physical mechanism and a correlation analysis method, automatically learns and judges the incidence relation of the characteristic events by machine learning, and effectively improves the accuracy of event relation identification. And matching and judging the upper and lower level incidence relation of the characteristic event by constructing a two-level knowledge map. And identifying a cascading failure evolution path based on a depth-first search strategy according to the characteristic event incidence relation judgment result and the time sequence characteristic, and meeting the requirements of online tracing analysis and advanced prediction of a cascading failure chain. The method utilizes knowledge graph technology to replace thinking and judgment of scheduling personnel, and provides effective technical support for tracing and prejudging the difficult problems of complex cascading failures.
Example 2
In a second aspect, the invention relates to a cascading failure evolution path tracing and predicting device based on a knowledge graph, which comprises:
the data acquisition module is used for acquiring cascading failure historical data and generating a cascading failure evolution sample database;
the model training module is used for training to obtain a characteristic event relation model based on the obtained cascading failure evolution sample database, wherein the input of the characteristic event relation model is the key characteristic quantity of an event pair, and the output of the characteristic event relation model is whether the relation between the event pairs is established or not;
the knowledge graph building module is used for extracting event types, characteristic events, relations among the characteristic events and relation attributes from a cascading failure evolution sample database to form a characteristic event evolution knowledge graph, wherein the relations among the characteristic events are obtained by inputting a trained characteristic event relation model by any characteristic event composition event;
and the tracing prediction module is used for searching the characteristic event evolution knowledge graph aiming at the obtained actually measured characteristic event sequence to obtain the tracing and prediction path of the cascading failure characteristic event.
The specific implementation scheme of each module in the device of the invention refers to the implementation process of each step in the method.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application 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 application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. 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.
The above description is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, several modifications and variations can be made without departing from the technical principle of the present invention, and these modifications and variations should also be regarded as the protection scope of the present invention.
Claims (8)
1. A cascading failure evolution path tracing and predicting method based on a knowledge graph is characterized by comprising the following processes:
acquiring cascading failure historical data and generating a cascading failure evolution sample database;
training to obtain a characteristic event relation model based on the obtained cascading failure evolution sample database, wherein the input of the characteristic event relation model is the key characteristic quantity of an event pair, and the output of the characteristic event relation model is whether the relation between the event pair is established or not;
extracting event classes, characteristic events, relations among the characteristic events and relation attributes from a cascading failure evolution sample database to form a characteristic event evolution knowledge graph, wherein the relations among the characteristic events are obtained by inputting any characteristic event composition event into a trained characteristic event relation model;
and searching the characteristic event evolution knowledge graph aiming at the obtained actually measured characteristic event sequence to obtain the tracing and prediction path of the cascading failure characteristic event.
2. The method for tracing and predicting cascading failure evolution paths based on the knowledge graph as claimed in claim 1, wherein the cascading failure history data comprises event sequences arranged according to an event occurrence sequence, power grid operation mode data before and after an event, power grid transient electrical quantity change information caused by the event, a failure location and a failure type.
3. The method for tracing and predicting cascading failure evolution paths based on the knowledge graph as claimed in claim 1, wherein the training process of the characteristic event relation model comprises:
determining a training event pair set from a cascading failure evolution sample database through physical mechanism analysis;
extracting key characteristic quantity of each event pair set by a data correlation analysis method;
and training based on a support vector machine algorithm by taking the key characteristic quantity as input and whether the relation between the event pairs is established as output to obtain a characteristic event relation model.
4. The method for tracing and predicting cascading failure evolution paths based on the knowledge graph as claimed in claim 1, wherein the extracting of event classes, characteristic events, relationships among the characteristic events and key characteristic quantities to form the characteristic event evolution knowledge graph comprises:
taking the event class as a node in the top layer;
taking the specific characteristic event as a bottom layer node, and connecting the specific characteristic event with the event class belonging to the upper layer node by using a connecting line;
in the bottom node, if the events have a relationship, connecting the two events in the graph by using an arrow connecting line, and pointing the arrow to a subsequent event;
and taking the relation attribute of the event as a node led out by the bottom node.
5. The method of claim 1, wherein the relationship between the characteristic events is derived by regularized knowledge.
6. The method for tracing and predicting cascading failure evolution path based on the knowledge graph as claimed in claim 1, wherein the step of searching the feature event evolution knowledge graph for the obtained actually measured feature event sequence to obtain the tracing and predicting path of the cascading failure feature event comprises the steps of:
arranging the actually measured characteristic event sequence according to a time sequence to form a characteristic event set;
removing feature events which are not matched with the feature event evolution knowledge graph from the feature event set;
searching all related characteristic events upwards by adopting a depth-first search algorithm and taking the sequenced last characteristic event in the characteristic event set as a starting point;
forming event pairs by all searched characteristic events according to a search sequence, inputting the event pairs into a characteristic event relation model to obtain the relation between the characteristic events, and if the relation between the characteristic events is 'true' and the events exist in a characteristic event set, bringing the event pairs into a tracing path set;
sequencing the characteristic events in the tracing path set according to a time sequence, wherein the sequenced event sequence is the tracing path of the cascading failure;
a depth-first search algorithm is adopted, the last sequenced characteristic event in the characteristic event set is taken as a starting point, and all the related characteristic events are searched downwards;
forming event pairs by all searched characteristic events according to a search sequence, inputting the event pairs into a characteristic event relation model to obtain the relation between the characteristic events, and if the relation between the characteristic events is 'true' and the events exist in a characteristic event set, bringing the event pairs into a prediction path set;
and the event sequence in the predicted path set is the predicted cascading failure evolution path.
7. The method for tracing and predicting cascading failure evolution paths based on the knowledge graph as claimed in claim 1, wherein the step of eliminating the feature events which do not have matching in the feature event evolution knowledge graph from the feature event set comprises:
preferentially searching whether a matched event class exists in the feature event evolution knowledge graph or not aiming at each feature event in the feature event set, and if not, removing the feature event from the feature event set;
and if the event class to which the characteristic event belongs exists, matching the key characteristic quantity of the event characteristic with the relationship attributes of the event characteristics and the event characteristics in the event class aiming at the event characteristic, and if the matching is unsuccessful, removing the characteristic event from the characteristic event set.
8. A cascading failure evolution path tracing and predicting device based on a knowledge graph is characterized by comprising the following steps:
the data acquisition module is used for acquiring cascading failure historical data and generating a cascading failure evolution sample database;
the model training module is used for training to obtain a characteristic event relation model based on the obtained cascading failure evolution sample database, wherein the input of the characteristic event relation model is the key characteristic quantity of an event pair, and the output of the characteristic event relation model is whether the relation between the event pairs is established or not;
the knowledge graph building module is used for extracting event types, characteristic events, relations among the characteristic events and relation attributes from a cascading failure evolution sample database to form a characteristic event evolution knowledge graph, wherein the relations among the characteristic events are obtained by inputting a trained characteristic event relation model by any characteristic event composition event;
and the tracing prediction module is used for searching the characteristic event evolution knowledge graph aiming at the obtained actually measured characteristic event sequence to obtain the tracing and prediction path of the cascading failure characteristic event.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110191955.XA CN112990551A (en) | 2021-02-20 | 2021-02-20 | Cascading failure evolution path tracing and predicting method and device based on knowledge graph |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110191955.XA CN112990551A (en) | 2021-02-20 | 2021-02-20 | Cascading failure evolution path tracing and predicting method and device based on knowledge graph |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112990551A true CN112990551A (en) | 2021-06-18 |
Family
ID=76393605
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110191955.XA Withdrawn CN112990551A (en) | 2021-02-20 | 2021-02-20 | Cascading failure evolution path tracing and predicting method and device based on knowledge graph |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112990551A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113705874A (en) * | 2021-08-19 | 2021-11-26 | 国家电网有限公司 | New energy power grid evolution prediction method and device, computer equipment and storage medium |
| CN114493408A (en) * | 2021-12-21 | 2022-05-13 | 重庆特斯联智慧科技股份有限公司 | Knowledge graph-based cargo optimized loading method and device |
| CN115081918A (en) * | 2022-07-07 | 2022-09-20 | 北京交通大学 | Rail transit risk point prediction method and system based on data driving |
| WO2023282738A1 (en) * | 2021-07-09 | 2023-01-12 | Universiti Tenaga Nasional | A computer-implemented method and system for modeling and predicting failure of a power grid configuration |
| CN117192373A (en) * | 2023-08-08 | 2023-12-08 | 浙江凌骁能源科技有限公司 | Power battery fault analysis method, device, computer equipment and storage medium |
| CN118014564A (en) * | 2024-04-10 | 2024-05-10 | 山东和兑智能科技有限公司 | Power equipment fault diagnosis system and method based on data driving |
-
2021
- 2021-02-20 CN CN202110191955.XA patent/CN112990551A/en not_active Withdrawn
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023282738A1 (en) * | 2021-07-09 | 2023-01-12 | Universiti Tenaga Nasional | A computer-implemented method and system for modeling and predicting failure of a power grid configuration |
| CN113705874A (en) * | 2021-08-19 | 2021-11-26 | 国家电网有限公司 | New energy power grid evolution prediction method and device, computer equipment and storage medium |
| CN113705874B (en) * | 2021-08-19 | 2024-03-15 | 国家电网有限公司 | New energy power grid evolution prediction method, device, computer equipment and storage medium |
| CN114493408A (en) * | 2021-12-21 | 2022-05-13 | 重庆特斯联智慧科技股份有限公司 | Knowledge graph-based cargo optimized loading method and device |
| CN114493408B (en) * | 2021-12-21 | 2024-04-09 | 重庆特斯联智慧科技股份有限公司 | Goods optimized loading method and device based on knowledge graph |
| CN115081918A (en) * | 2022-07-07 | 2022-09-20 | 北京交通大学 | Rail transit risk point prediction method and system based on data driving |
| CN115081918B (en) * | 2022-07-07 | 2023-05-23 | 北京交通大学 | Rail transit risk point prediction method and system based on data driving |
| CN117192373A (en) * | 2023-08-08 | 2023-12-08 | 浙江凌骁能源科技有限公司 | Power battery fault analysis method, device, computer equipment and storage medium |
| CN117192373B (en) * | 2023-08-08 | 2024-05-07 | 浙江凌骁能源科技有限公司 | Power battery fault analysis method, device, computer equipment and storage medium |
| CN118014564A (en) * | 2024-04-10 | 2024-05-10 | 山东和兑智能科技有限公司 | Power equipment fault diagnosis system and method based on data driving |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112990551A (en) | Cascading failure evolution path tracing and predicting method and device based on knowledge graph | |
| CN107274105B (en) | Linear discriminant analysis-based multi-attribute decision tree power grid stability margin evaluation method | |
| CN109501834B (en) | Method and device for predicting fault of turnout switch machine | |
| CN111985653B (en) | Power grid fault knowledge recommendation and knowledge management system and method based on knowledge graph | |
| CN113283602B (en) | Offline knowledge learning and online dynamic correction-based power grid system fault diagnosis method | |
| CN103197168B (en) | Realize the method for fault diagnosis control based on the event set chain of causation in electric system | |
| CN116304928A (en) | Power supply equipment fault prediction method, device, equipment and storage medium | |
| CN110968703B (en) | Method and system for constructing abnormal metering point knowledge base based on LSTM end-to-end extraction algorithm | |
| CN111049131B (en) | Regional power grid online fault handling plan generation method and system | |
| Wang et al. | Data-driven transient stability assessment using sparse PMU sampling and online self-check function | |
| CN107103361A (en) | Diagnosis Method of Transformer Faults and system based on rough set and rebound strength curve | |
| CN116227878B (en) | Power load decomposition analysis method, system, computer device and storage medium | |
| Xu et al. | Fault diagnosis and identification of malfunctioning protection devices in a power system via time series similarity matching | |
| CN117034149A (en) | Fault processing strategy determining method and device, electronic equipment and storage medium | |
| Jiarula et al. | Fault mode prediction based on decision tree | |
| CN113837423A (en) | Power grid operation situation prediction method based on energy internet electric power big data | |
| Kinghorst et al. | Integration of additional information sources for improved alarm flood detection | |
| Olaru et al. | Data mining tools and application in power system engineering | |
| Shaaban et al. | Transformer fault diagnosis method based on rough set and generalized distribution table | |
| CN114399177A (en) | Scheduling disposal rule mining and generating method and system based on Apriori | |
| Chhokra et al. | A binary decision diagram based cascade prognostics scheme for power systems | |
| Tao et al. | Power Grid Fault Diagnosis Method Based on LightGBM | |
| Tan et al. | Automatic Construction of Knowledge Graph and Its Application in Electric Power System | |
| CN117872038B (en) | DC micro-grid instability fault source positioning method and device based on graph theory | |
| Huang et al. | Power Grid Fault Diagnosis Based on Random Forest |
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 | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20210618 |
|
| WW01 | Invention patent application withdrawn after publication |