CN115239033B - Method for generating causal model under corresponding power grid operation environment - Google Patents
Method for generating causal model under corresponding power grid operation environment Download PDFInfo
- Publication number
- CN115239033B CN115239033B CN202211169268.9A CN202211169268A CN115239033B CN 115239033 B CN115239033 B CN 115239033B CN 202211169268 A CN202211169268 A CN 202211169268A CN 115239033 B CN115239033 B CN 115239033B
- Authority
- CN
- China
- Prior art keywords
- event
- power grid
- entities
- causal
- entity
- 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.)
- Active
Links
- 230000001364 causal effect Effects 0.000 title claims abstract description 160
- 238000000034 method Methods 0.000 title claims abstract description 42
- 238000013528 artificial neural network Methods 0.000 claims abstract description 43
- 238000005259 measurement Methods 0.000 claims abstract description 22
- 238000004364 calculation method Methods 0.000 claims abstract description 16
- 238000012216 screening Methods 0.000 claims abstract description 10
- 238000007621 cluster analysis Methods 0.000 claims abstract description 7
- 239000006185 dispersion Substances 0.000 claims description 35
- 238000004458 analytical method Methods 0.000 claims description 34
- 238000013527 convolutional neural network Methods 0.000 claims description 24
- 238000012549 training Methods 0.000 claims description 10
- 238000004422 calculation algorithm Methods 0.000 claims description 6
- 230000003252 repetitive effect Effects 0.000 claims description 5
- 238000012935 Averaging Methods 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 4
- 238000011161 development Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 241000845077 Iare Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010219 correlation analysis Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000013024 troubleshooting 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
- G06N20/00—Machine learning
- G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/02—Knowledge representation; Symbolic representation
- G06N5/022—Knowledge engineering; Knowledge acquisition
- G06N5/025—Extracting rules from data
-
- 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
-
- 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)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Economics (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Human Resources & Organizations (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Strategic Management (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Operations Research (AREA)
- Game Theory and Decision Science (AREA)
- Medical Informatics (AREA)
- Development Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Quality & Reliability (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a method for generating a causal model under a corresponding power grid operating environment, which comprises the following steps: the cluster analysis is utilized to carry out combined screening on the power grid equipment entities based on the operation condition data to obtain power grid equipment entities with various power grid event types as the swept entities of the various power grid event types; and constructing an event entity correlation measurement and calculation model representing the event correlation between waves and entities in each power grid event category by utilizing a neural network, and measuring and calculating the event correlation between the waves and the entities based on the event entity correlation measurement and calculation model. According to the method, the event entity causal model between the medium waves and the entities in each power grid event category is constructed, so that the prediction of the causal relationship between the medium waves and the entities in the power grid event category is realized, a large number of causal relationships among power grid equipment can be visually displayed, the discovery efficiency of event sources by utilizing the causal relationships is improved, and the complexity of an event tracing process is reduced.
Description
Technical Field
The invention relates to the technical field of power grid analysis, in particular to a method for generating a causal model in a corresponding power grid operating environment.
Background
With the continuous development of scientific technology and the continuous improvement of power systems, the system not only brings infinite convenience (such as illumination) to the life of people, but also provides more convenience (such as electric vehicles, trains, trams and the like) in the aspects of travel and the like of people. And the comprehensive speed-up of the construction of the extra-high voltage power grid, the rapid development of new energy and the deep promotion of electric power market reformation are realized, the characteristics of the integrated operation of the power grid are more obvious, and the requirements of centralized analysis and decision-making, multi-professional business cooperation, cross-dispatching mechanism work cooperation and the like on the power grid are more urgent.
In recent years, the information resources which can be acquired by a power grid dispatching control center are more and more abundant, more and more information sources can be used for regulating and analyzing power grid equipment, and the information quantity is huge, such as the regulation and control big data of the power grid equipment. However, the big data statistical study itself is a statistical analysis method, and the cause-and-effect relationship on logic cannot be checked, and the cause-and-effect relationship of the power grid equipment with defect events cannot be intuitively reflected, so that an observer cannot intuitively find the cause-and-effect relationship among the equipment from a large number of power grid equipment, and further, the cause-and-effect relationship is difficult to find the source of the event, the event tracing process is complicated, and the efficiency is low.
Disclosure of Invention
The invention aims to provide a method for generating a causal model corresponding to a power grid operating environment, and the method is used for solving the technical problems that in the prior art, an observer cannot intuitively find causal relationships among equipment from a large number of power grid equipment, further the causal relationships are difficult to find event sources, the event tracing process is complicated, and the efficiency is low.
In order to solve the technical problems, the invention specifically provides the following technical scheme:
a method for generating a causal model in a corresponding power grid operating environment comprises the following steps:
the method comprises the following steps that S1, operation condition data of each power grid equipment entity when each power grid event type occurs are sequentially extracted from a power grid log, and the power grid equipment entities subjected to combined screening based on the operation condition data by utilizing cluster analysis are used for obtaining power grid equipment entities subjected to various power grid event types as various power grid event types;
s2, constructing an event entity correlation measurement and calculation model representing event correlation between waves and entities in each power grid event type by utilizing a neural network, and measuring and calculating the event correlation between the waves and the entities based on the event entity correlation measurement and calculation model;
and S3, carrying out discrete analysis and mean analysis on the event correlation among the waves and the entities in each power grid event category in sequence to determine the causal relationship among the waves and the entities in each power grid event category, and constructing an event entity causal model among the waves and the entities in each power grid event category based on the causal relationship among the waves and the entities in each power grid event category and the event data of each power grid event category so as to realize the prediction of the causal relationship among the waves and the entities in the power grid event category.
As a preferred scheme of the present invention, the method for obtaining a power grid equipment entity affected by each power grid event category by performing combination screening on the power grid equipment entities based on operation condition data by using cluster analysis as an affected entity of each power grid event category includes:
in each power grid event category, clustering each power grid equipment entity into a plurality of entity sets by using a Kmeans clustering algorithm based on operation condition data, and screening out the power grid equipment entity closest to a clustering center of each entity set as a center entity;
inputting the operation condition data of each central entity into a pre-established event type prediction model to obtain the occurrence event type of each central entity, sequentially comparing the occurrence event type of each central entity with the power grid event type, wherein,
if the occurrence event type of the central entity is consistent with the power grid event type, all power grid equipment entities in the entity set to which the corresponding central entity belongs are marked as swept entities of the power grid event type to obtain the swept entities of each power grid event type;
if the occurrence event type of the central entity is inconsistent with the power grid event type, all power grid equipment entities in the entity set of the corresponding central entity are marked as non-swept entities of the power grid event type;
the construction of the event category prediction model comprises the following steps:
the method comprises the following steps of taking operation condition data of each power grid equipment entity when each power grid event type occurs as a CNN neural network input item, taking a power grid event type corresponding to the operation condition data as a CNN neural network output item, and performing model training on the CNN neural network input item and the CNN neural network output item by using the CNN neural network to obtain an event type prediction model of each power grid equipment entity, wherein the model expression of the event type prediction model is as follows:
Label i =CNN(data i )；
in the formula, label i Is as followsiEvent class, data, of individual grid equipment entities i Is as followsiThe operation condition data of each power grid equipment entity, CNN is CNN neural network,iare the metering constants.
As a preferable scheme of the invention, the operation condition data is normalized before the Kmeans clustering algorithm operation and the event type prediction model operation.
As a preferred embodiment of the present invention, the constructing an event entity correlation measurement model characterizing event correlations between waves and entities in each grid event category by using a neural network includes:
selecting a plurality of swept entities as sample entities in each power grid event category, and extracting operating condition data of the sample entities when the power grid event category occurs as event data of the sample entities;
sequentially calculating the data similarity of the event data of any two sample entities, taking the event data and the power grid event categories of the sample entities as BP neural network input items, taking the data similarity between the sample entities as BP neural network output items, and performing model training on the BP neural network input items and the BP neural network output items by utilizing a BP neural network to obtain a measurement and calculation model of the event entity correlation in each power grid event category;
the model expression of the event entity correlation measurement model is as follows:
P=BP(D j ,D k ,Label)；
wherein P is the event correlation between entities, D j ,D k Are respectively the firstj，kThe event data of each entity, label is the category of the power grid event, BP is a BP neural network,j，kis a metering constant.
As a preferred embodiment of the present invention, the measuring and calculating the event correlation between the swept entities based on the event entity correlation measurement model includes:
combining all the swept entities in each power grid event category into the event entity correlation measurement model in a pairwise manner to obtain event correlation of each combination in each swept entity in each power grid event category;
extracting repeated swept entities from each power grid event category as repeated entities, and taking event correlation of pairwise combination in the repeated swept entities as event correlation of pairwise combination in the repeated entities.
As a preferred aspect of the present invention, the sequentially performing discrete analysis and averaging analysis on the event correlations between the waves and the entities in each power grid event category to determine the causal relationship between the waves and the entities in each power grid event category includes:
discrete analysis is carried out on event correlation in each power grid event category combined in pairs in the repeated entities to obtain repeated entity correlation dispersion, the repeated entity correlation dispersion is compared with preset dispersion, wherein,
if the correlation dispersion of the repeated entities is less than or equal to the preset dispersion, marking the pairwise combination of the corresponding repeated entities as a causal combination;
if the correlation dispersion of the repeated entities is greater than the preset dispersion, marking the pairwise combination of the corresponding repeated entities as a non-causal combination;
carrying out averaging analysis on event correlation of the causal combinations in each power grid event category to obtain a causal combination correlation mean value, and comparing the causal combination correlation mean value with a preset mean value, wherein,
if the causal combination correlation mean value is larger than or equal to the preset mean value, marking the corresponding causal combination as a strong causal combination;
if the correlation mean value of the causal combinations is smaller than the preset mean value, marking the corresponding causal combinations as weak causal combinations;
and marking the causal relationship between the affected entities corresponding to the strong causal combination as a strong causal relationship, marking the causal relationship between the affected entities corresponding to the weak causal combination as a weak causal relationship, and marking the causal relationship between the affected entities except the strong causal combination and the weak causal combination as a random causal relationship.
As a preferred aspect of the present invention, the constructing a causal model of event entities between waves and entities in each power grid event category based on the causal relationship between waves and entities in each power grid event category and event data in each power grid event category includes:
taking event data among the swept entities in each power grid event category as an input item of an SVM classifier, taking causal relationships among the swept entities as output items of the SVM classifier, and performing model training by using the SVM classifier based on the input item and the output item of the SVM classifier to obtain an event entity causal model, wherein a model expression of the event entity causal model is as follows:
Type=SVM([data x _list],[data y _list])；
wherein Type is the cause and effect relationship between entities, [ data ] x _list],[data y _list]The data sequences are respectively represented by the data sequences formed by the event data of the x-th and y-th swept entities in each power grid event category, and the SVM is an SVM classifier.
As a preferred embodiment of the present invention, the repetitive entity correlation dispersion is measured by using variance, and the calculation formula of the repetitive entity correlation dispersion is as follows:
wherein V is the dispersion of the correlation of the repeated entities, P lr Is as followslEvent correlation of the first and the r-th repeating entities,lr is a metering constant, n is the total number of repeated entities;
the calculation formula of the causal combination correlation mean value is as follows:
where aver is the mean of causal combinatory correlations, P hg For the h repeated entity and the h repeated entity marked as causal combinationsgEvent correlation of the individual duplicate entities, h,gfor the metric constant, m is the total number of duplicate entities labeled as causal combinations.
As a preferable aspect of the present invention, the data similarity is calculated using a euclidean distance, a cosine similarity, or a correlation coefficient.
As a preferred scheme of the present invention, the operation condition data is obtained in a power grid operation environment.
Compared with the prior art, the invention has the following beneficial effects:
according to the method, discrete analysis and mean analysis are sequentially carried out on event correlation among the waves and the entities in each power grid event category to determine the causal relationship among the waves and the entities in each power grid event category, and an event entity causal model among the waves and the entities in each power grid event category is constructed on the basis of the causal relationship among the waves and the entities in each power grid event category and event data of each power grid event category, so that the causal relationship among the waves and the entities in the power grid event category is predicted, a large number of causal relationships among power grid equipment can be visually displayed, the discovery efficiency of event sources by using the causal relationships is improved, and the complexity of an event tracing process is reduced.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below. It should be apparent that the drawings in the following description are merely exemplary, and that other embodiments can be derived from the drawings provided by those of ordinary skill in the art without inventive effort.
FIG. 1 is a flow chart of a method for a big data causal model according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
As shown in fig. 1, the present invention provides a method for generating a causal model in a corresponding power grid operating environment, comprising the following steps:
the method comprises the following steps that S1, operation condition data of each power grid equipment entity when each power grid event type occurs are sequentially extracted from a power grid log, and the power grid equipment entities subjected to combined screening based on the operation condition data by utilizing cluster analysis are used for obtaining power grid equipment entities subjected to various power grid event types as various power grid event types;
the method for obtaining the power grid equipment entity affected by each power grid event type by utilizing cluster analysis to carry out combined screening on the power grid equipment entity based on the operation condition data as the affected entity of each power grid event type comprises the following steps:
in each power grid event category, clustering each power grid equipment entity into a plurality of entity sets by using a Kmeans clustering algorithm based on operation condition data, and screening out a power grid equipment entity which is closest to a clustering center of each entity set as a center entity;
inputting the operation condition data of each central entity into a pre-established event type prediction model to obtain the occurrence event type of each central entity, sequentially comparing the occurrence event type of each central entity with the event type of the power grid, wherein,
if the occurrence event type of the central entity is consistent with the power grid event type, all power grid equipment entities in the entity set to which the corresponding central entity belongs are marked as swept entities of the power grid event type to obtain the swept entities of each power grid event type;
if the occurrence event type of the central entity is inconsistent with the power grid event type, all power grid equipment entities in the entity set of the corresponding central entity are marked as non-swept entities of the power grid event type;
the construction of the event category prediction model comprises the following steps:
the method comprises the following steps of taking operation condition data of each power grid equipment entity when each power grid event type occurs as a CNN neural network input item, taking a power grid event type corresponding to the operation condition data as a CNN neural network output item, and performing model training on the CNN neural network input item and the CNN neural network output item by using the CNN neural network to obtain an event type prediction model of each power grid equipment entity, wherein the model expression of the event type prediction model is as follows:
Label i =CNN(data i )；
in the formula, label i Is as followsiEvent class, data, of individual grid equipment entities i Is as followsiThe operation condition data of each power grid equipment entity, CNN is CNN neural network,ifor measuringAnd (4) counting.
And normalizing the operation condition data before performing Kmeans clustering algorithm operation and event type prediction model operation.
In the power grid event analysis under the real scene, the occurrence source of the event is usually checked, that is, the equipment entity serving as the occurrence source is checked, so that the causal relationship among the equipment entities is identified in the power grid event analysis, and the event source can be checked according to the causal relationship of the equipment entities.
In this embodiment, each grid event category is analyzed one by one, so as to analyze the grid device entity, i.e., the swept entity in this embodiment, in each grid event category, where the grid event category occurs, so as to obtain a grid device entity having a causal relationship preliminarily obtained in the grid event category, which is used as a basis for subsequent analysis of the causal relationship.
S2, constructing an event entity correlation measurement and calculation model representing event correlation between waves and entities in each power grid event type by utilizing a neural network, and measuring and calculating the event correlation between the waves and the entities based on the event entity correlation measurement and calculation model;
an event entity correlation measurement model for representing event correlation between waves and entities in each power grid event category is constructed by utilizing a neural network, and the method comprises the following steps:
selecting a plurality of swept entities as sample entities in each power grid event category, and extracting operating condition data of the sample entities when the power grid event category occurs as event data of the sample entities;
sequentially calculating the data similarity of the event data of any two sample entities, taking the event data and the power grid event categories of the sample entities as BP neural network input items, taking the data similarity between the sample entities as BP neural network output items, and performing model training on the BP neural network input items and the BP neural network output items by utilizing a BP neural network to obtain an event entity correlation measurement model in each power grid event category;
the model expression of the event entity correlation measurement model is as follows:
P=BP(D j ,D k ,Label)；
wherein P is the event correlation between entities, D j ,D k Are respectively the firstj，kThe event data of each entity, label is the category of the power grid event, BP is a BP neural network,j，kis a metering constant.
The method for measuring and calculating the event correlation between the spread entities based on the event entity correlation measuring and calculating model comprises the following steps:
combining all the swept entities in each power grid event category into an event entity correlation measurement model in a pairwise manner to obtain event correlation of each swept entity in each power grid event category in the pairwise manner;
extracting repeated swept-over entities from each power grid event category as repeated entities, and taking event correlation of two-by-two combination of the repeated swept-over entities as event correlation of two-by-two combination of the repeated entities, wherein if the swept-over entities in the power grid event category a include a power grid equipment entity 1, a power grid equipment entity 2 and a power grid equipment entity 6, and if the swept-over entities in the power grid event category B include a power grid equipment entity 1, a power grid equipment entity 2, a power grid equipment entity 3, a power grid equipment entity 6 and a power grid equipment entity 7, and if the swept-over entities in the power grid event category C include a power grid equipment entity 1, a power grid equipment entity 2, a power grid equipment entity 4, a power grid equipment entity 6 and a power grid equipment entity 9, then selecting the repeated power grid equipment entities 1, the power grid equipment entity 2 and the power grid equipment entity 6 in the power grid event categories, and indicating that event equipment entities occur among the three power grid events, and concluding that the relationship among the three power grid equipment entities exists, so that the relationship can be determined.
The causal relationship is equivalent to the correlation relationship in a broad sense, so that the causal relationship of the power grid equipment entity is determined by analyzing the correlation of the power grid equipment entity, and the automatic quantitative analysis of the correlation relationship is realized by constructing an event entity correlation measurement and calculation model, so that the correlation analysis efficiency and accuracy are improved.
And S3, carrying out discrete analysis and mean analysis on the event correlation among the waves and the entities in each power grid event category in sequence to determine the causal relationship among the waves and the entities in each power grid event category, and constructing an event entity causal model among the waves and the entities in each power grid event category based on the causal relationship among the waves and the entities in each power grid event category and the event data of each power grid event category so as to realize the prediction of the causal relationship among the waves and the entities in the power grid event category.
The event correlation of the waves and the entities in each power grid event category is subjected to discrete analysis and mean analysis in sequence to determine the causal relationship between the waves and the entities in each power grid event category, and the method comprises the following steps:
discrete analysis is carried out on event correlation in each power grid event category combined in pairs in the repeated entities to obtain repeated entity correlation dispersion, the repeated entity correlation dispersion is compared with preset dispersion, wherein,
if the correlation dispersion of the repeated entities is less than or equal to the preset dispersion, marking the pairwise combination of the corresponding repeated entities as a causal combination;
if the correlation dispersion of the repeated entities is greater than the preset dispersion, marking the pairwise combination of the corresponding repeated entities as a non-causal combination;
averaging and analyzing the event correlation of the causal combinations in each power grid event category to obtain a causal combination correlation mean value, and comparing the causal combination correlation mean value with a preset mean value, wherein,
if the causal combination correlation mean value is greater than or equal to a preset mean value, marking the corresponding causal combination as a strong causal combination;
if the causal combination correlation mean value is smaller than a preset mean value, marking the corresponding causal combination as a weak causal combination;
and marking the causal relationship between the affected entities corresponding to the strong causal combination as a strong causal relationship, marking the causal relationship between the affected entities corresponding to the weak causal combination as a weak causal relationship, and marking the causal relationship between the affected entities except the strong causal combination and the weak causal combination as a random causal relationship.
The correlation dispersion of the repeated entity is measured by using variance, and the calculation formula of the correlation dispersion of the repeated entity is as follows:
wherein V is the dispersion of the correlation of the repeated entities, P lr Is a firstlEvent correlation of the first and the r-th repeating entities,lr is a metering constant, n is the total number of repeated entities;
the calculation formula of the causal combination correlation mean value is as follows:
where aver is the mean of causal combinatory correlations, P hg For the h repeated entity and the h repeated entity marked as causal combinationsgThe event correlation of the individual repeated entities, h,gfor the metric constant, m is the total number of duplicate entities labeled as causal combinations.
The higher the correlation dispersion, the more stable the correlation between the grid equipment entities in each grid event category is, so that a stable correlation relationship exists between the grid equipment entities, and then through analysis of the correlation mean, the higher the correlation dispersion and the higher the correlation mean, the more stable the correlation between the grid equipment entities is, and the stronger the correlation is, so that this embodiment takes this stable strong correlation as a strong causal relationship, for example: the event correlation between the grid equipment entity 1 and the grid equipment entity 2 in the grid event categories a, B and C is 0.8, 0.9 and 0.85, the correlation dispersion and the correlation mean value of the grid equipment entity 1 and the grid equipment entity 2 are both high, so the grid equipment entity 1 and the grid equipment entity 2 are strong causal relationship, the event correlation between the grid equipment entity 1 and the grid equipment entity 6 in the grid event categories a, B and C is 0.2, 0.1 and 0.15, the correlation dispersion and the correlation mean value of the grid equipment entity 1 and the grid equipment entity 2 are both low, so the grid equipment entity 1 and the grid equipment entity 2 are weak causal relationship.
When detecting that the power grid events A, B and C occur in the power grid equipment entity 2 in the subsequent time sequence, the traceability value of the power grid equipment entity 1 can be obtained, traceability can be rapidly realized according to the causal relationship, the power grid equipment entity affected by the events can be rapidly mastered, the fault troubleshooting time is shortened, the timeliness of detection and maintenance is stronger, and the equipment loss is reduced.
Constructing an event entity causal model between the medium waves and the entities of each power grid event type based on the causal relationship between the medium waves and the entities of each power grid event type and the event data of each power grid event type, wherein the event entity causal model comprises the following steps:
taking event data in each power grid event category among the swept entities as an SVM classifier input item, taking causal relationships among the swept entities as an SVM classifier output item, performing model training by using an SVM classifier based on the SVM classifier input item and the SVM classifier output item to obtain an event entity causal model, wherein the model expression of the event entity causal model is as follows:
Type=SVM([data x _list],[data y _list])；
wherein Type is the cause and effect relationship between entities, [ data ] x _list],[data y _list]The data sequences are respectively represented by the data sequences formed by the event data of the x-th and y-th swept entities in each power grid event category, and the SVM is an SVM classifier.
The causal relationship and the event data are subjected to data rule summarization to construct an event entity causal model, wherein training samples of the event entity causal model come from the causal relationship determined by discrete analysis and mean analysis, so that the prediction result of the event entity causal model can accord with the analysis result of the discrete analysis and the mean analysis, namely, the event entity causal model performs convolution learning on discrete steps of the discrete analysis, the mean analysis and the like, the causal relationship between waves and entities in each power grid event category can be directly obtained from the event data of each power grid event category, and the operation simplicity is improved.
The data similarity is calculated using euclidean distance, cosine similarity or correlation coefficient.
And acquiring the operating condition data in the power grid operating environment.
According to the method, discrete analysis and mean analysis are sequentially carried out on event correlation among waves and entities in each power grid event category to determine the causal relationship among the waves and the entities in each power grid event category, and an event entity causal model among the waves and the entities in each power grid event category is constructed on the basis of the causal relationship among the waves and the entities in each power grid event category and event data of each power grid event category, so that the causal relationship among the waves and the entities in the power grid event category can be predicted, a large number of causal relationships among power grid equipment can be visually displayed, the efficiency of finding event sources by using the causal relationship is improved, and the complexity of an event tracing process is reduced.
The above embodiments are only exemplary embodiments of the present application, and are not intended to limit the present application, and the protection scope of the present application is defined by the claims. Various modifications and equivalents may be made by those skilled in the art within the spirit and scope of the present application and such modifications and equivalents should also be considered to be within the scope of the present application.
Claims (10)
1. A method for generating a causal model corresponding to a power grid operating environment is characterized in that: the method comprises the following steps:
s1, sequentially extracting operation condition data of each power grid equipment entity when each power grid event type occurs from a power grid log, and performing combined screening on the power grid equipment entities by using cluster analysis based on the operation condition data to obtain power grid equipment entities affected by each power grid event type as affected entities of each power grid event type;
s2, constructing an event entity correlation measurement and calculation model representing event correlation between waves and entities in each power grid event category by utilizing a neural network, and measuring and calculating the event correlation between the waves and the entities based on the event entity correlation measurement and calculation model;
and S3, carrying out discrete analysis and mean analysis on the event correlation among the waves and the entities in each power grid event category in sequence to determine the causal relationship among the waves and the entities in each power grid event category, and constructing an event entity causal model among the waves and the entities in each power grid event category based on the causal relationship among the waves and the entities in each power grid event category and the event data of each power grid event category so as to realize the prediction of the causal relationship among the waves and the entities in the power grid event category.
2. Method for generating a causal model for a corresponding grid operating environment according to claim 1, wherein: the method for utilizing cluster analysis to carry out combined screening on the power grid equipment entities based on the operation condition data to obtain the power grid equipment entities spread by each power grid event category as the spread entities of each power grid event category comprises the following steps:
in each power grid event category, clustering each power grid equipment entity into a plurality of entity sets by using a Kmeans clustering algorithm based on operation condition data, and screening out the power grid equipment entity closest to a clustering center of each entity set as a center entity;
inputting the operation condition data of each central entity into a pre-established event type prediction model to obtain the occurrence event type of each central entity, sequentially comparing the occurrence event type of each central entity with the event type of the power grid, wherein,
if the occurrence event type of the central entity is consistent with the power grid event type, all power grid equipment entities in the entity set to which the corresponding central entity belongs are marked as swept entities of the power grid event type to obtain the swept entities of each power grid event type;
if the occurrence event type of the central entity is inconsistent with the power grid event type, all power grid equipment entities in the entity set of the corresponding central entity are marked as non-swept entities of the power grid event type;
the construction of the event category prediction model comprises the following steps:
the method comprises the following steps of taking operation condition data of each power grid equipment entity when each power grid event type occurs as a CNN neural network input item, taking a power grid event type corresponding to the operation condition data as a CNN neural network output item, and performing model training on the CNN neural network input item and the CNN neural network output item by using the CNN neural network to obtain an event type prediction model of each power grid equipment entity, wherein the model expression of the event type prediction model is as follows:
Label i =CNN(data i )；
in the formula, label i Is a firstiEvent class, data, of individual grid equipment entities i Is as followsiThe operation condition data of each power grid equipment entity, CNN is CNN neural network,iis a metering constant.
3. The method for generating a causal model for a grid operating environment according to claim 2, wherein: and normalizing the operation condition data before performing Kmeans clustering algorithm operation and event type prediction model operation.
4. A method for generating a causal model for a corresponding grid operating environment according to claim 3, wherein: the method for constructing the event entity correlation measurement model for representing the event correlation between waves and entities in each power grid event category by utilizing the neural network comprises the following steps of:
selecting a plurality of swept entities as sample entities in each power grid event category, and extracting operating condition data of the sample entities when the power grid event categories occur as event data of the sample entities;
sequentially calculating the data similarity of the event data of any two sample entities, taking the event data of the sample entities and the power grid event category as BP neural network input items, taking the data similarity between the sample entities as BP neural network output items, and performing model training on the BP neural network input items and the BP neural network output items by using the BP neural network to obtain a measurement and calculation model of the event entity correlation in each power grid event category;
the model expression of the event entity correlation measurement model is as follows:
P=BP(D j ,D k ,Label)；
wherein P is the event correlation between entities, D j ,D k Are respectively the firstj，kEvent data of each entity, label is a power grid event category, BP is a BP neural network,j，kare the metering constants.
5. Method for generating a causal model for a corresponding grid operating environment according to claim 4, wherein: the measuring and calculating of the event correlation between the affected entities based on the event entity correlation measuring and calculating model comprises the following steps:
combining all the swept entities in each power grid event category into the event entity correlation measurement model in pairs to obtain event correlation of each swept entity in each power grid event category in pairs;
extracting repeated swept entities from each power grid event category as repeated entities, and taking event correlation of pairwise combination in the repeated swept entities as event correlation of pairwise combination in the repeated entities.
6. Method for generating a causal model for a corresponding grid operating environment according to claim 5, wherein: the method for sequentially performing discrete analysis and mean analysis on the event correlation among the waves and the entities in each power grid event category to determine the causal relationship among the waves and the entities in each power grid event category comprises the following steps:
performing dispersion analysis on event correlation of each duplicate entity in each power grid event category to obtain duplicate entity correlation dispersion, and comparing the duplicate entity correlation dispersion with a preset dispersion, wherein,
if the correlation dispersion of the repeated entities is less than or equal to the preset dispersion, marking the pairwise combination of the corresponding repeated entities as a causal combination;
if the correlation dispersion of the repeated entities is greater than the preset dispersion, marking the pairwise combination of the corresponding repeated entities as a non-causal combination;
carrying out averaging analysis on event correlation of the causal combinations in each power grid event category to obtain a causal combination correlation mean value, and comparing the causal combination correlation mean value with a preset mean value, wherein,
if the causal combination correlation mean value is greater than or equal to a preset mean value, marking the corresponding causal combination as a strong causal combination;
if the correlation mean value of the causal combinations is smaller than the preset mean value, marking the corresponding causal combinations as weak causal combinations;
and marking the causal relationship between the affected entities corresponding to the strong causal combination as a strong causal relationship, marking the causal relationship between the affected entities corresponding to the weak causal combination as a weak causal relationship, and marking the causal relationship between the affected entities except the strong causal combination and the weak causal combination as a random causal relationship.
7. The method for generating a causal model for a grid operating environment as claimed in claim 6, wherein said constructing a causal model for event entities in each grid event category based on causal relationships between events and entities in each grid event category and event data for each grid event category comprises:
taking event data among the swept entities in each power grid event category as an input item of an SVM classifier, taking causal relationships among the swept entities as output items of the SVM classifier, and performing model training by using the SVM classifier based on the input item and the output item of the SVM classifier to obtain an event entity causal model, wherein a model expression of the event entity causal model is as follows:
Type=SVM([data x _list],[data y _list])；
wherein Type is the causal relationship between entities, [ data ] x _list],[data y _list]The data sequences are respectively represented by event data of the x-th and y-th swept entities in each power grid event category, and the SVM is an SVM classifier.
8. The method for generating a causal model for an operating environment of a corresponding power grid as claimed in claim 7, wherein said repetitive entity correlation dispersion is measured using variance, and said repetitive entity correlation dispersion is calculated by the formula:
wherein V is the dispersion of the correlation of the repetitive entities, P lr Is a firstlEvent correlation of the individual repeating entities and the r-th repeating entity,lr is a metering constant, n is the total number of repeated entities;
the calculation formula of the causal combination correlation mean value is as follows:
wherein aver is the mean of causal combined correlations, P hg For the h-th repeated entity and the h-th repeated entity marked as causal combinationsgEvent correlation of the individual duplicate entities, h,gto measure the constant, m is the total number of repeated entities marked as causal combinations.
9. The method for generating a causal model under a corresponding power grid operating environment of claim 4, wherein the data similarity is calculated using Euclidean distance, cosine similarity, or correlation coefficient.
10. The method for generating a causal model for a grid operating environment of claim 2, wherein the operating condition data is obtained from the grid operating environment.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211169268.9A CN115239033B (en) | 2022-09-26 | 2022-09-26 | Method for generating causal model under corresponding power grid operation environment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211169268.9A CN115239033B (en) | 2022-09-26 | 2022-09-26 | Method for generating causal model under corresponding power grid operation environment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115239033A CN115239033A (en) | 2022-10-25 |
| CN115239033B true CN115239033B (en) | 2023-02-10 |
Family
ID=83667283
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211169268.9A Active CN115239033B (en) | 2022-09-26 | 2022-09-26 | Method for generating causal model under corresponding power grid operation environment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115239033B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109754158A (en) * | 2018-12-07 | 2019-05-14 | 国网江苏省电力有限公司南京供电分公司 | A method of generating the big data Causal model under corresponding operation of power networks environment |
| CN112910089A (en) * | 2021-01-25 | 2021-06-04 | 国网山东省电力公司青岛供电公司 | Transformer substation secondary equipment fault logic visualization method and system |
| CN113268590A (en) * | 2021-04-06 | 2021-08-17 | 云南电网有限责任公司昆明供电局 | Power grid equipment running state evaluation method based on equipment portrait and integrated learning |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP5364530B2 (en) * | 2009-10-09 | 2013-12-11 | 株式会社日立製作所 | Equipment state monitoring method, monitoring system, and monitoring program |
| CN103105556B (en) * | 2013-01-30 | 2015-01-07 | 西安交通大学 | Intelligent power grid load testing and recognition method based on steady state and transient state characteristic joint matching |
| CN103514516B (en) * | 2013-09-27 | 2015-04-29 | 国家电网公司 | Method for obtaining power grid device failure information based on calculation of multi-source redundant information |
-
2022
- 2022-09-26 CN CN202211169268.9A patent/CN115239033B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109754158A (en) * | 2018-12-07 | 2019-05-14 | 国网江苏省电力有限公司南京供电分公司 | A method of generating the big data Causal model under corresponding operation of power networks environment |
| CN112910089A (en) * | 2021-01-25 | 2021-06-04 | 国网山东省电力公司青岛供电公司 | Transformer substation secondary equipment fault logic visualization method and system |
| CN113268590A (en) * | 2021-04-06 | 2021-08-17 | 云南电网有限责任公司昆明供电局 | Power grid equipment running state evaluation method based on equipment portrait and integrated learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115239033A (en) | 2022-10-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Thamarai et al. | House Price Prediction Modeling Using Machine Learning. | |
| Sanjeev | Implementation of machine learning algorithms for analysis and prediction of air quality | |
| CN110703057A (en) | Power equipment partial discharge diagnosis method based on data enhancement and neural network | |
| CN111900731A (en) | PMU-based power system state estimation performance evaluation method | |
| CN117131449A (en) | Data management-oriented anomaly identification method and system with propagation learning capability | |
| CN111551888A (en) | Electric energy meter metering data fault analysis method based on improved AdaBoost algorithm | |
| CN112181706A (en) | Power dispatching data anomaly detection method based on logarithmic interval isolation | |
| WO2021103401A1 (en) | Data object classification method and apparatus, computer device and storage medium | |
| CN114416423B (en) | Root cause positioning method and system based on machine learning | |
| CN115564194A (en) | Method and system for constructing metering abnormality diagnosis information generation model of smart power grid | |
| CN115373879A (en) | Intelligent operation and maintenance disk fault prediction method for large-scale cloud data center | |
| Wang et al. | R2-trans: Fine-grained visual categorization with redundancy reduction | |
| CN117035697B (en) | ITSM (integrated traffic simulation) platform optimization method and system based on historical dynamic analysis | |
| Quintana et al. | ALDI++: Automatic and parameter-less discord and outlier detection for building energy load profiles | |
| CN115239033B (en) | Method for generating causal model under corresponding power grid operation environment | |
| CN111737993A (en) | Method for extracting health state of equipment from fault defect text of power distribution network equipment | |
| CN116720118A (en) | Label quality intelligent analysis method and device, electronic equipment and storage medium | |
| CN115828156A (en) | Power grid metadata monitoring-based electricity stealing and leakage identification method and system | |
| CN114139408A (en) | Power transformer health state assessment method | |
| CN115209441A (en) | Method, device, equipment and storage medium for predicting base station out-of-service alarm | |
| CN112085421A (en) | Teacher teaching effect evaluation method based on deep BP neural network | |
| Yang et al. | Differentiated analysis for music traffic in software defined networks: A method of deep learning | |
| CN112884167B (en) | Multi-index anomaly detection method based on machine learning and application system thereof | |
| CN111565192A (en) | Credibility-based multi-model cooperative defense method for internal network security threats | |
| Glushkovsky | Ai giving back to statistics? discovery of the coordinate system of univariate distributions by beta variational autoencoder |
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 | ||
| GR01 | Patent grant |