CN117290404A - Method and system for rapidly searching and practical main distribution network fault processing method - Google Patents
Method and system for rapidly searching and practical main distribution network fault processing method Download PDFInfo
- Publication number
- CN117290404A CN117290404A CN202311107886.5A CN202311107886A CN117290404A CN 117290404 A CN117290404 A CN 117290404A CN 202311107886 A CN202311107886 A CN 202311107886A CN 117290404 A CN117290404 A CN 117290404A
- Authority
- CN
- China
- Prior art keywords
- processing
- database
- case
- vector
- distribution network
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 64
- 238000009826 distribution Methods 0.000 title claims abstract description 37
- 238000003672 processing method Methods 0.000 title claims abstract description 27
- 238000012545 processing Methods 0.000 claims abstract description 83
- 230000000694 effects Effects 0.000 claims abstract description 20
- 238000004458 analytical method Methods 0.000 claims abstract description 13
- 230000008569 process Effects 0.000 claims abstract description 13
- 238000005457 optimization Methods 0.000 claims abstract description 11
- 238000012163 sequencing technique Methods 0.000 claims abstract description 6
- 239000013598 vector Substances 0.000 claims description 71
- 238000011156 evaluation Methods 0.000 claims description 24
- 238000000605 extraction Methods 0.000 claims description 15
- 238000004590 computer program Methods 0.000 claims description 11
- 230000007547 defect Effects 0.000 claims description 11
- 230000007246 mechanism Effects 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 7
- 238000003860 storage Methods 0.000 claims description 7
- 238000011282 treatment Methods 0.000 claims description 7
- 230000004913 activation Effects 0.000 claims description 6
- 238000010276 construction Methods 0.000 claims description 6
- 230000006870 function Effects 0.000 claims description 6
- 230000011218 segmentation Effects 0.000 claims description 6
- 238000013528 artificial neural network Methods 0.000 claims description 4
- 238000004422 calculation algorithm Methods 0.000 claims description 4
- 238000002372 labelling Methods 0.000 claims description 4
- 230000008092 positive effect Effects 0.000 claims description 4
- 238000012360 testing method Methods 0.000 claims description 4
- 238000012512 characterization method Methods 0.000 claims description 3
- 238000001514 detection method Methods 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 210000002569 neuron Anatomy 0.000 claims description 3
- 238000012549 training Methods 0.000 claims description 3
- 230000001939 inductive effect Effects 0.000 claims description 2
- 230000010354 integration Effects 0.000 claims 1
- 238000012986 modification Methods 0.000 claims 1
- 230000004048 modification Effects 0.000 claims 1
- 230000006872 improvement Effects 0.000 abstract description 3
- 238000004891 communication Methods 0.000 description 4
- 238000012423 maintenance Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000012937 correction Methods 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000004904 shortening Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2474—Sequence data queries, e.g. querying versioned data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/21—Design, administration or maintenance of databases
- G06F16/211—Schema design and management
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/21—Design, administration or maintenance of databases
- G06F16/215—Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/242—Query formulation
- G06F16/243—Natural language query formulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2457—Query processing with adaptation to user needs
- G06F16/24578—Query processing with adaptation to user needs using ranking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0499—Feedforward networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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
- G06Q10/20—Administration of product repair or maintenance
-
- 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)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Databases & Information Systems (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Business, Economics & Management (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Business, Economics & Management (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Quality & Reliability (AREA)
- Marketing (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- Evolutionary Biology (AREA)
- Operations Research (AREA)
- Entrepreneurship & Innovation (AREA)
- Probability & Statistics with Applications (AREA)
- Fuzzy Systems (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Primary Health Care (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a practical rapid retrieval method of a main distribution network fault processing method, which relates to the technical field of rapid retrieval and comprises the steps of collecting historical problem processing records and constructing a problem processing database; constructing a fault retrieval and matching system based on a database; outputting an analysis process and a processing measure of the matching case as a reference scheme of the current fault; and feeding back the processing result, and continuously optimizing iteration. The invention can accelerate fault processing speed, provide reliable reference schemes, promote sequencing accuracy of processing schemes, predict processing effects, provide diversified processing schemes, and continuously improve the performance and practicability of the whole system through continuous optimization and improvement.
Description
Technical Field
The invention relates to the technical field of quick retrieval, in particular to a practical quick retrieval method and a system for a main distribution network fault processing method.
Background
At present, some enterprises or organizations establish databases for processing faults of a main distribution network, collect a great deal of history problem processing records and arrange the records, partially utilize text matching, pattern matching and other technologies, search similar cases from the databases according to fault descriptions input by users, match and recommend the similar cases, and partially utilize natural language processing, text analysis and other technologies for analyzing and processing the fault descriptions, extract key information and assist in fault searching and matching.
However, since the database establishment process may involve inconsistent participation of multiple persons and data collection, quality problems of the database, such as missing key information, inaccurate records, etc., are caused; in addition, the existing fault searching and matching method may have the problem of low precision, and a processing scheme similar to the current fault and effective to the current fault cannot be accurately found, so that the recommended scheme is not accurate and reliable enough. Therefore, the development of rapid retrieval of the main distribution network fault processing has advanced to a certain extent, but the defects and challenges in the aspects of database quality, fault retrieval precision, information extraction completeness and the like still exist.
Disclosure of Invention
The invention is provided in view of the problem that the fault searching and matching method possibly has low precision, and the problem that the recommended scheme is inaccurate and reliable because the processing scheme similar to the current fault cannot be accurately found.
Therefore, the problem to be solved by the present invention is how to provide a method for accelerating fault handling, providing reliable reference schemes, improving the sequencing accuracy of handling schemes, predicting handling effects, and providing diversified handling schemes.
In order to solve the technical problems, the invention provides the following technical scheme:
in a first aspect, an embodiment of the present invention provides a method for quickly searching for a failure processing method of a main distribution network, which includes collecting a history problem processing record, and constructing a problem processing database; constructing a fault retrieval and matching system based on a database; outputting an analysis process and a processing measure of the matching case as a reference scheme of the current fault; and feeding back the processing result, and continuously optimizing iteration.
As a preferable scheme of the main distribution network fault processing method of the invention, a practical method for fast searching is adopted, wherein: the process of constructing the problem handling database is as follows: collecting historical problem records; extracting text key information; constructing a structured database; evaluating the quality of the knowledge database; the specific steps for evaluating the quality of the database are as follows: randomly extracting n samples from a database as an evaluation data set, wherein the sample size n is set to be 100 or more, extracting entities on the evaluation data set, comparing the entity extraction with manually marked entity results, and taking the correct proportion of the identified entities as the accuracy rate; based on the extracted entity, automatically generating structural data by using a template, comparing the structural data with a manual labeling result, and calculating the data duty ratio meeting the requirement to be used as a filling satisfaction rate; if the entity extraction accuracy is lower than the set value of 85%, the entity identification model needs to be optimized; if the filling satisfaction rate is lower than 80% of the set value, the structure of the problem template needs to be adjusted.
As a preferable scheme of the main distribution network fault processing method of the invention, a practical method for fast searching is adopted, wherein: the process for constructing the fault retrieval and matching system based on the database comprises the steps of constructing a case feature vector, inputting query vectorization representation, carrying out vector similarity comparison, and ranking similar cases; the vector similarity comparison includes, assuming that an input query vector is a and a case vector is B, cosine similarity is:
Q＝cos(θ)＝AxB/(||A||*||B||)
wherein θ is the angle between the two vectors, axB is the vector inner product, A and B are the vector length, the cosine value range is [ -1,1], the larger the value is, the more similar; using vector dimension to represent the feature quantity of word vector, wherein the higher the dimension is, the richer the semantic information of the characterization words is; dividing the vector space of the corpus into M areas, wherein each area contains N vectors, N < < N, selecting one vector as a root node, calculating the distance between other vectors and the root node, wherein the distance is greater than a median at a far-end node and less than a median at a near-end node, and recursively constructing subtrees until leaf nodes; dividing the super sphere by taking the root node as the sphere center and the median distance, wherein the sphere radius of the near-terminal node is the median distance, and the far-terminal node is arranged in the outer sphere area; and calculating the distance d between the query vector and the root node, searching the near-end subtree if d is smaller than the median, searching the far-end subtree if d is greater than the median, and narrowing the searching range layer by layer until the leaf nodes.
As a preferable scheme of the main distribution network fault processing method of the invention, a practical method for fast searching is adopted, wherein: the ranking of similar cases comprises the steps of collecting the cases marked by the positive and negative surfaces with the processed effect as model training data, segmenting the text by using a word segmentation algorithm, obtaining all words in the text, and calculating by using the following formula:
TF-IDF＝TF*IDF
wherein TF is the number of times of occurrence of words in the text/total word number; IDF is log text total number/word appearance text number; setting the size M of word vectors, wherein each word corresponds to a vector with the length M, taking a TF-IDF value as one dimension of the word vector, adopting a feedforward neural network structure, receiving case feature vectors by an input layer, performing feature learning by a hidden layer, and extracting hidden features related to sequencing; the output layer sets a node to represent the sorting score, and the node is initialized to a fully-connected structure; primary judgment: whether to output linearly, if so, using a ReLU activation function to perform nonlinear mapping, and if not, maintaining linear output; judging whether an attention mechanism is needed secondarily, judging whether the importance of different features to the ordering is different, if so, adding attention sub-network re-weighting features, and if so, directly and fully connecting and outputting; the model inputs case characteristics, outputs a real number score, obtains a required sorting score by setting a range and a nonlinear function, and the larger the score is, the more similar the query case is.
As a preferable scheme of the main distribution network fault processing method of the invention, a practical method for fast searching is adopted, wherein: the ranking of similar cases further comprises adding a neuron to the output layer, acquiring 0-1 prediction probability by using sigmoid activation, predicting positive effects if the probability is greater than or equal to 0.5, and predicting negative effects if the probability is less than 0.5.
As a preferable scheme of the main distribution network fault processing method of the invention, a practical method for fast searching is adopted, wherein: the analysis process and the processing measures of the output matching cases comprise judging whether a case which is identical to the current fault exists or not, and if so, directly outputting the processing details of the case; if no case is absolutely the same, calculating the approximation degree of the current fault and the database case, if the approximation degree is more than or equal to K, the approximation degree is regarded as an approximation case, and if the approximation degree is less than K, the approximation degree is regarded as a reference case; and selecting the case with the similarity score of 5 at the top of the rank from the approximate cases, appropriately modifying the schemes of the similar cases to generate a new processing scheme, carrying out multi-scheme combination or expert correction, and processing the situations which are not absolutely identical.
As a preferable scheme of the main distribution network fault processing method of the invention, a practical method for fast searching is adopted, wherein: the continuous optimization iterative process comprises the steps of defining objective evaluation indexes simultaneously when a processing scheme is given, wherein the evaluation indexes are obtained based on state detection without manual judgment; after each time period T of implementing the processing scheme, the system automatically collects the defined evaluation index, and compares the evaluation index value with a predefined expected target value; if the deviation exceeds the threshold value, judging that the processing effect is poor, automatically starting root cause analysis, analyzing the cause of the deviation, inducing database defects or scheme defects which cause the deviation, optimizing links such as case feature extraction and representation in the database, adjusting a scoring mechanism of a model, compensating the defects of a generated scheme, retraining the model and continuously testing until an evaluation index reaches the standard.
In a second aspect, in order to further solve the problem that the fault searching and matching method may have low precision, and the problem that the recommended scheme is inaccurate and reliable because the similar and effective processing scheme cannot be found accurately, the embodiment provides a fast searching and practical system of the main distribution network fault processing method, which comprises a problem processing database construction module, a fault processing database processing module and a fault processing database processing module, wherein the problem processing database construction module is used for collecting and constructing a structured problem processing database; the feature expression and vectorization module is used for realizing feature vectorization expression of the case; the similarity calculation and sorting module is used for realizing the calculation and sorting of the similarities among the cases, ranking the cases according to the similarities and outputting the most matched cases; the processing scheme generating module is used for giving a processing scheme according to the matching case; and the evaluation and optimization module is used for completing processing feedback, iterating and optimizing the system, improving the system knowledge perfection degree and matching accuracy.
In a third aspect, embodiments of the present invention provide a computer apparatus comprising a memory and a processor, the memory storing a computer program, wherein: the computer program when executed by a processor implements any step of the main distribution network fault handling method according to the first aspect of the present invention for fast retrieval of a practical method.
In a fourth aspect, embodiments of the present invention provide a computer-readable storage medium having a computer program stored thereon, wherein: the computer program when executed by a processor implements any step of the main distribution network fault handling method according to the first aspect of the present invention for fast retrieval of a practical method.
The invention has the beneficial effects that the invention develops a retrieval tool for the main distribution network fault and the processing method, assists operation and maintenance personnel in solving the problem, reduces the workload of dispatching operation and maintenance personnel, and improves the stability and reliability of a power grid system; the tool is an offline tool, does not depend on a network and a system, has no safety problem and is convenient to popularize; the invention can accelerate fault processing speed, provide reliable reference schemes, promote sequencing accuracy of processing schemes, predict processing effects, provide diversified processing schemes, and continuously improve the performance and practicability of the whole system through continuous optimization and improvement.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art. Wherein:
fig. 1 is a flowchart of a fast search utility method for a failure processing method of a main distribution network in embodiment 1.
Fig. 2 is an operation schematic diagram of a fast search utility method of the main distribution network fault handling method in embodiment 1.
Detailed Description
In order that the above-recited objects, features and advantages of the present invention will become more readily apparent, a more particular description of the invention will be rendered by reference to specific embodiments thereof which are illustrated in the appended drawings.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, but the present invention may be practiced in other ways other than those described herein, and persons skilled in the art will readily appreciate that the present invention is not limited to the specific embodiments disclosed below.
Further, reference herein to "one embodiment" or "an embodiment" means that a particular feature, structure, or characteristic can be included in at least one implementation of the invention. The appearances of the phrase "in one embodiment" in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments.
Example 1
Referring to fig. 1 and 2, a first embodiment of the present invention provides a method for quickly searching for a failure handling method of a main distribution network, which includes the following steps:
s1: and collecting historical problem processing records and constructing a problem processing database.
S1.1: historical problem records are collected.
Further, a problem collection template is set, and the problem collection template comprises fields such as problem description, treatment measures, treatment effects and the like; the historical problem records are derived and different problem types, such as software faults, hardware faults or network faults, are distinguished.
S1.2: and extracting text key information.
Preferably, the problem description text is subjected to preprocessing such as word segmentation, part-of-speech tagging and the like; extracting keywords and entities related to the problem by using named entity recognition; judging and marking whether the treatment effect is positive or negative.
S1.3: a structured database is constructed.
Defining a structured template for different keyword types, and filling each field of the template according to the extracted information; if the extraction effect of a certain type of problem is poor, the key information is manually supplemented.
S1.4: database quality is assessed.
Evaluating the database sample, and calculating indexes such as entity extraction accuracy, filling satisfaction rate and the like; if the index is lower than the threshold, the template and the extraction algorithm need to be adjusted; and iteratively optimizing the quality of the database to meet the application requirement.
Further, randomly extracting n samples from the database as an evaluation data set, wherein the sample size n can be set to be 100 or more, extracting the entity on the evaluation data set, comparing with the manually marked entity result, and calculating the correct proportion in the identified entity as the accuracy; based on the extracted entity, automatically generating structural data by using a template, comparing the structural data with a manual labeling result, and calculating the data duty ratio meeting the requirement to be used as a filling satisfaction rate; if the entity extraction accuracy is lower than 85%, the entity identification model needs to be optimized; if the filling satisfaction rate is lower than 80%, the structure of the problem template needs to be adjusted.
S2: and constructing a fault retrieval and matching system based on the database.
S2.1: a case feature vector is constructed and a query vectorized representation is input.
Extracting information such as text, codes, logs and the like for each historical case, and constructing text feature vectors in a word vector mode and the like; and word segmentation and vectorization are carried out on the input problem description text.
S2.2: and (5) vector similarity comparison is carried out.
Preferably, the calculating the similarity between the input query vector and the case vector by using the vector space model specifically includes: let the input query vector be a, and a case vector be B, then the cosine similarity is:
Q＝cos(θ)＝AxB/(||A||*||B||)
where θ is the angle between the two vectors, axB is the vector inner product, A and B are the vector length, the cosine value range is [ -1,1], and larger values represent more similar.
The vector dimension represents the feature quantity of word vectors, and the higher the dimension is, the richer the semantic information of the characterization words is; for example, a 300-dimensional word vector may represent word semantic relationships that are more accurate than simple word frequency statistics.
VP tree index construction is selected, and the method specifically comprises the following steps: dividing the vector space of the corpus into M areas, wherein each area contains N vectors, N < < N, selecting one vector as a root node, calculating the distance between other vectors and the root node, wherein the distance is greater than a median at a far-end node and less than a median at a near-end node, and recursively constructing subtrees until leaf nodes; dividing the super sphere by taking the root node as the sphere center and the median distance, wherein the sphere radius of the near-terminal node is the median distance, and the far-terminal node is arranged in the outer sphere area; and calculating the distance d between the query vector and the root node, searching the near-end subtree if d is smaller than the median, searching the far-end subtree if d is greater than the median, and narrowing the searching range layer by layer until the leaf nodes.
It should be noted that, the branches are gradually filtered by using vector distance comparison, the distance between the query and the corpus vector is not required to be calculated, and the actual distance calculation is only performed on the related branches, so that the complexity is reduced, and the purpose of accelerating the search is achieved.
S2.3: similar cases were ranked.
Collecting the cases marked by the positive and negative surfaces with the processing effect as model training data, segmenting the text by using a word segmentation algorithm, acquiring all words in the text, and calculating by using the following formula:
TF-IDF＝TF*IDF
wherein TF is the number of times of occurrence of words in the text/total word number; IDF is log (total number of text/number of text occurrences); setting the dimension of a word vector, for example, 300, wherein each word corresponds to a vector with the length of 300, taking the TF-IDF value as one dimension of the word vector, and if n words are obtained by word segmentation of a text, the length of the word vector corresponding to each word is 300, and splicing the n word vectors to obtain a text feature vector with the length of 300 x n; and the feed-forward neural network structure is adopted, the input layer receives the case feature vector, the hidden layer performs feature learning, and the hidden features related to the ordering are extracted.
Further, the output layer sets a node to represent the sorting score, the node is initialized to a fully connected structure, and the primary judgment is as follows: whether to output linearly, if so, using a ReLU activation function to perform nonlinear mapping, and if not, maintaining linear output; judging whether an attention mechanism is needed secondarily, judging whether the importance of different features to the ordering is different, if so, adding attention sub-network re-weighting features, and if so, directly and fully connecting and outputting; the model inputs case characteristics, outputs a real number score, obtains a required sorting score by setting a range and a nonlinear function, and the larger the score is, the more similar the query case is.
Furthermore, a neuron is added in the output layer, the sigmoid activation is used for obtaining 0-1 prediction probability, if the probability is more than or equal to 0.5, the positive effect is predicted, and if the probability is less than the positive effect, the negative effect is predicted.
S3: and outputting the analysis process and the processing measures of the matching cases as a reference scheme of the current faults.
Preferably, judging whether a case which is identical with the current fault exists or not, if so, directly outputting the processing details of the case; if the fault is not the same case, calculating the approximation degree of the current fault and the database case, if the approximation degree is more than or equal to 85%, the fault is regarded as an approximation case, and if the fault is less than 85%, the fault is regarded as a reference case;
the top 5 cases of the approximation score rank are selected from the approximation cases, the schemes of the approximation cases are appropriately modified to generate new processing schemes, multiple scheme combination or expert correction may be needed to process the situations which are not absolutely identical.
S4: and feeding back the processing result, and continuously optimizing iteration.
Further, when a processing scheme is given, objective evaluation indexes are defined at the same time, the indexes can be obtained based on log analysis, state detection and the like, manual judgment is not needed, after the processing scheme is implemented for a period of time, a system automatically collects the defined evaluation indexes, compares the index values with a predefined expected target value, if the deviation exceeds a threshold value, the processing effect is judged to be poor, root cause analysis is automatically started, the reasons for the deviation are analyzed, database defects or scheme defects which cause the deviation are induced, links such as case feature extraction and representation in a database are optimized, a scoring mechanism of a model is adjusted, the defects of a generated scheme are compensated, the model is retrained and the test is continued until the evaluation indexes reach the standard.
The embodiment also provides a rapid retrieval practical system of the main distribution network fault processing method, which comprises a problem processing database construction module, a database processing module and a database processing module, wherein the problem processing database construction module is used for collecting and constructing a structured problem processing database; the feature expression and vectorization module is used for realizing feature vectorization expression of the case; the similarity calculation and sorting module is used for realizing the calculation and sorting of the similarities among the cases, ranking the cases according to the similarities and outputting the most matched cases; the processing scheme generating module is used for giving a processing scheme according to the matching case; and the evaluation and optimization module is used for completing processing feedback, iterating and optimizing the system, improving the system knowledge perfection degree and matching accuracy.
The embodiment also provides a computer device, which is suitable for the situation of fast searching a practical method by using the main distribution network fault processing method, and comprises the following steps: a memory and a processor; the memory is used for storing computer executable instructions, and the processor is used for executing the computer executable instructions to realize the rapid retrieval practical method of the main distribution network fault processing method according to the embodiment; the industrial control device comprises an FPGA and a memory, wherein the memory stores an FPGA program.
The computer device may be a terminal comprising a processor, a memory, a communication interface, a display screen and input means connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The communication interface of the computer device is used for carrying out wired or wireless communication with an external terminal, and the wireless mode can be realized through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer equipment can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer equipment can be a touch layer covered on the display screen, can also be keys, a track ball or a touch pad arranged on the shell of the computer equipment, and can also be an external keyboard, a touch pad or a mouse and the like.
The present embodiment also provides a storage medium having stored thereon a computer program which, when executed by a processor, implements a practical method for implementing a fast retrieval of a failure handling method for a main distribution network as proposed in the above embodiment; the storage medium may be implemented by any type or combination of volatile or nonvolatile Memory devices, such as static random access Memory (Static Random Access Memory, SRAM), electrically erasable Programmable Read-Only Memory (Electrically Erasable Programmable Read-Only Memory, EEPROM), erasable Programmable Read-Only Memory (Erasable Programmable Read Only Memory, EPROM), programmable Read-Only Memory (PROM), read-Only Memory (ROM), magnetic Memory, flash Memory, magnetic disk, or optical disk.
In summary, the invention develops a retrieval tool of a main distribution network fault and processing method, assists operation and maintenance personnel in processing problems, reduces the workload of dispatching operation and maintenance personnel, and improves the stability and reliability of a power grid system; the tool is an offline tool, does not depend on a network and a system, has no safety problem and is convenient to popularize; the invention can accelerate fault processing speed, provide reliable reference schemes, promote sequencing accuracy of processing schemes, predict processing effects, provide diversified processing schemes, and continuously improve the performance and practicability of the whole system through continuous optimization and improvement.
Example 2
Referring to table 1, in order to verify the beneficial effects of the second embodiment of the present invention, based on the first embodiment, experimental simulation data of a fast retrieval practical method of the main distribution network fault processing method of the present invention is provided.
Firstly, collecting historical fault records, extracting text key information, and constructing a structured database. The specific steps for evaluating database quality are as follows: randomly extracting 100 samples from the database as an evaluation data set; and (3) extracting the entity on the evaluation data set, comparing the result with the manual annotation, and calculating the accuracy.
Generating structured data by using templates based on the extraction entity, comparing the structured data with the manual labeling result, and calculating to generate a satisfaction rate; if the entity extraction accuracy rate is less than 85%, optimizing an entity identification model; if the satisfaction rate is less than 80%, the problem template structure is adjusted.
Wherein, the partial fault table is as follows:
TABLE 1 Fault template Table
| Fields | Description of the invention |
| Fault numbering | Uniquely identifying fault records |
| Time to failure | Time of failure occurrence |
| Fault apparatus | Name of the failed device |
| Description of faults | Literal description of fault conditions |
| Cause of failure | Root cause of failure |
| Treatment measures | Specific measures taken to solve the fault |
| Processing the result | Fault handling results are classified into good and bad |
The following table is a comparison of the present invention with the prior art:
table 2 comparative table of the inventive and prior art treatments
It can be seen that the table compares the performances of the invention with the prior art on a plurality of key indexes, and from the scale of a case library, the invention collects more than 5000 cases, which is more than 4 times of 1200 cases in the prior art, and fully utilizes the advantage of large data accumulation; in the aspect of case extraction accuracy, the method reaches 95.89%, and the method improves the accuracy by 6.35%. This means that key information can be accurately identified from large-scale cases; in the aspect of the success rate of search matching, the invention realizes 92.3% of successful matching, which exceeds 4.66 percentage points in the prior art. The vector space model retrieval mechanism constructed by the invention has obvious effect; the accuracy of the processing scheme is a key index for evaluating the actual effect. The invention reaches 87.09% on the index, and the effect is improved by 11.4% points; the invention realizes the rapid iterative optimization once a week, and the prior art needs one month, thereby greatly shortening the optimization period; in terms of fault handling time length, the average time length of the method is 2.13 hours, which is shortened by nearly 40% compared with 3.41 hours in the prior art, so that the power grid fault response speed is directly improved.
In summary, the system and the method have the advantages that the system and the method are remarkably improved in multiple dimensions such as case scale, accuracy, response speed and the like, the rapid and accurate processing of power grid faults can be effectively supported, and the stability and reliability of the system are improved.
Constructing a case feature vector, inputting a query to carry out vectorization, calculating cosine similarity between the query vector and the case vector, finding out similar cases, and ranking and outputting the similar cases; performing similar case scoring by using a neural network, and according to the output ranking score, the larger the score is, the more similar the score is to the query case, and querying the fault processing method; after the treatment is implemented for a period of time, automatically collecting index values, comparing the index values with target values, if the deviation exceeds a threshold value, judging that the effect is poor, starting root cause analysis, analyzing database or model defects which cause the deviation, carrying out optimization adjustment, retraining the model, and testing the index until reaching the standard.
According to the invention, by constructing the problem processing database and establishing the fault retrieval and matching mechanism based on the database, the processing scheme of the historical similar faults can be quickly found out, and a reference is provided for current fault analysis. Simultaneously, the iterative database and the matching model are continuously optimized by combining the result feedback, and the quality of the processing scheme is continuously improved.
It should be noted that the above embodiments are only for illustrating the technical solution of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention may be modified or substituted without departing from the spirit and scope of the technical solution of the present invention, which is intended to be covered in the scope of the claims of the present invention.
Claims (10)
1. A practical method for quickly searching a main distribution network fault processing method is characterized in that: comprising the following steps:
collecting historical problem processing records and constructing a problem processing database;
constructing a fault retrieval and matching system based on a database;
outputting an analysis process and a processing measure of the matching case as a reference scheme of the current fault;
and feeding back the processing result, and continuously optimizing iteration.
2. The method for quickly retrieving and using the main distribution network fault handling method according to claim 1, wherein the method comprises the following steps: the process of constructing the problem handling database is as follows:
collecting historical problem records; extracting text key information; constructing a structured database; evaluating the quality of the knowledge database;
the specific steps for evaluating the quality of the database are as follows: randomly extracting n samples from a database to serve as an evaluation data set, setting a sample size n, extracting entities on the evaluation data set, comparing the entity extraction with manually marked entity results, and taking the correct proportion of identified entities as the accuracy;
based on the extracted entity, automatically generating structural data by using a template, comparing the structural data with a manual labeling result, and calculating the data duty ratio meeting the requirement to be used as a filling satisfaction rate; if the entity extraction accuracy is lower than the set value P, the entity identification model needs to be optimized; if the filling satisfaction rate is lower than the set value Q, the structure of the problem template needs to be adjusted.
3. The method for quickly retrieving and using the main distribution network fault handling method according to claim 2, wherein: the process for constructing the fault retrieval and matching system based on the database comprises the steps of constructing a case feature vector, inputting query vectorization representation, carrying out vector similarity comparison, and ranking similar cases;
the vector similarity comparison includes,
let the input query vector be a, and a case vector be B, then the cosine similarity is:
Q＝cos(θ)＝AxB/(||A||*||B||)
wherein θ is the angle between the two vectors, axB is the vector inner product, A and B are the vector length, the cosine value range is [ -1,1], the larger the value is, the more similar;
using vector dimension to represent the feature quantity of word vector, wherein the higher the dimension is, the richer the semantic information of the characterization words is;
dividing the vector space of the corpus into M areas, wherein each area contains N vectors, N < < N, selecting one vector as a root node, calculating the distance between other vectors and the root node, wherein the distance is greater than a median at a far-end node and less than a median at a near-end node, and recursively constructing subtrees until leaf nodes;
dividing the super sphere by taking the root node as the sphere center and the median distance, wherein the sphere radius of the near-terminal node is the median distance, and the far-terminal node is arranged in the outer sphere area;
and calculating the distance d between the query vector and the root node, searching the near-end subtree if d is smaller than the median, searching the far-end subtree if d is greater than the median, and narrowing the searching range layer by layer until the leaf nodes.
4. A method for quickly retrieving and utilizing a failure processing method of a main distribution network as set forth in claim 3, wherein: the ranking of similar cases includes,
collecting the case marked by the positive and negative surfaces of the processed effect as model training data, segmenting the text by using a word segmentation algorithm, acquiring all words in the text, and calculating by using the following formula:
TF-IDF＝TF*IDF
wherein TF is the number of times of occurrence of words in the text/total word number; IDF is log text total number/word appearance text number;
setting the size M of word vectors, wherein each word corresponds to a vector with the length M, taking a TF-IDF value as one dimension of the word vector, adopting a feedforward neural network structure, receiving case feature vectors by an input layer, performing feature learning by a hidden layer, and extracting hidden features related to sequencing; the output layer sets a node to represent the sorting score, and the node is initialized to a fully-connected structure;
primary judgment: whether to output linearly, if so, using a ReLU activation function to perform nonlinear mapping, and if not, maintaining linear output;
judging whether an attention mechanism is needed secondarily, judging whether the importance of different features to the ordering is different, if so, adding attention sub-network re-weighting features, and if so, directly and fully connecting and outputting;
the model inputs case characteristics, outputs a real number score, obtains a required sorting score by setting a range and a nonlinear function, and the larger the score is, the more similar the query case is.
5. The method for quickly retrieving and utilizing the failure processing method of the main distribution network as set forth in claim 4, wherein: the ranking of similar cases further includes,
and adding a neuron in the output layer, acquiring 0-1 prediction probability by using sigmoid activation, and predicting positive effects if the probability is more than or equal to 0.5, and predicting negative effects if the probability is less than 0.5.
6. The method for quickly retrieving and utilizing the failure processing method of the main distribution network according to claim 5, wherein the method comprises the following steps: the analysis process and processing means of the output matching cases include,
judging whether a case which is identical with the current fault exists or not, if so, directly outputting the processing details of the case;
if no case is absolutely the same, calculating the approximation degree of the current fault and the database case, if the approximation degree is more than or equal to K, the approximation degree is regarded as an approximation case, and if the approximation degree is less than K, the approximation degree is regarded as a reference case;
and selecting the case with the similarity score of 5 at the top of the ranking from the approximate cases, and carrying out integration modification on the schemes of the similar cases to generate a new processing scheme.
7. The method for quickly retrieving and utilizing the failure processing method of the main distribution network as set forth in claim 6, wherein: the process of continuing the optimization iteration includes,
when a treatment scheme is given, an objective evaluation index is defined at the same time, wherein the evaluation index is obtained based on state detection without manual judgment;
after each time period T of implementing the processing scheme, the system automatically collects the defined evaluation index, and compares the evaluation index value with a predefined expected target value;
if the deviation exceeds the threshold value, judging that the processing effect is poor, automatically starting root cause analysis, analyzing the cause of the deviation, inducing database defects or scheme defects which cause the deviation, optimizing links such as case feature extraction and representation in the database, adjusting a scoring mechanism of a model, compensating the defects of a generated scheme, retraining the model and continuously testing until an evaluation index reaches the standard.
8. A practical system for quickly retrieving a failure processing method of a main distribution network, which is based on the practical method for quickly retrieving the failure processing method of the main distribution network according to any one of claims 1 to 7, and is characterized in that: comprising the following steps:
the problem processing database construction module is used for collecting and constructing a structured problem processing database;
the feature expression and vectorization module is used for realizing feature vectorization expression of the case;
the similarity calculation and sorting module is used for realizing the calculation and sorting of the similarities among the cases, ranking the cases according to the similarities and outputting the most matched cases;
the processing scheme generating module is used for giving a processing scheme according to the matching case;
and the evaluation and optimization module is used for completing processing feedback, iterating and optimizing the system, improving the system knowledge perfection degree and matching accuracy.
9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that: the steps of the main distribution network fault handling method according to any one of claims 1 to 7 for fast retrieval of a practical method are implemented when the processor executes the computer program.
10. A computer-readable storage medium having stored thereon a computer program, characterized by: the steps of the main distribution network fault handling method according to any one of claims 1 to 7 for fast retrieval of a practical method are implemented when the computer program is executed by a processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311107886.5A CN117290404A (en) | 2023-08-30 | 2023-08-30 | Method and system for rapidly searching and practical main distribution network fault processing method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311107886.5A CN117290404A (en) | 2023-08-30 | 2023-08-30 | Method and system for rapidly searching and practical main distribution network fault processing method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117290404A true CN117290404A (en) | 2023-12-26 |
Family
ID=89247165
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311107886.5A Pending CN117290404A (en) | 2023-08-30 | 2023-08-30 | Method and system for rapidly searching and practical main distribution network fault processing method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117290404A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117556033A (en) * | 2024-01-11 | 2024-02-13 | 北京并行科技股份有限公司 | Method and device for determining embedded model parameters of question-answering system and computing equipment |
-
2023
- 2023-08-30 CN CN202311107886.5A patent/CN117290404A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117556033A (en) * | 2024-01-11 | 2024-02-13 | 北京并行科技股份有限公司 | Method and device for determining embedded model parameters of question-answering system and computing equipment |
| CN117556033B (en) * | 2024-01-11 | 2024-03-29 | 北京并行科技股份有限公司 | Method and device for determining embedded model parameters of question-answering system and computing equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US8126826B2 (en) | Method and system for active learning screening process with dynamic information modeling | |
| CN112632228A (en) | Text mining-based auxiliary bid evaluation method and system | |
| CN113779272A (en) | Data processing method, device and equipment based on knowledge graph and storage medium | |
| CN106054858A (en) | Decision tree classification and fault code classification-based vehicle remote diagnosis and spare part retrieval method | |
| CN111581545A (en) | Method for sorting recalled documents and related equipment | |
| CN111581956B (en) | Sensitive information identification method and system based on BERT model and K nearest neighbor | |
| CN111368096A (en) | Knowledge graph-based information analysis method, device, equipment and storage medium | |
| Thaler et al. | Towards a neural language model for signature extraction from forensic logs | |
| CN117290404A (en) | Method and system for rapidly searching and practical main distribution network fault processing method | |
| CN113449508B (en) | Internet public opinion correlation deduction prediction analysis method based on event chain | |
| Balaji et al. | Text summarization using NLP technique | |
| CN113505190A (en) | Address information correction method, device, computer equipment and storage medium | |
| Eskandari et al. | Predicting best answer using sentiment analysis in community question answering systems | |
| Saha et al. | A large scale study of SVM based methods for abstract screening in systematic reviews | |
| Oliveira et al. | Evolutionary rank aggregation for recommender systems | |
| CN116205296A (en) | ABAC strategy engineering method integrating top-down and bottom-up | |
| CN115269816A (en) | Core personnel mining method and device based on information processing method and storage medium | |
| Sushentsev et al. | Dapstep: Deep assignee prediction for stack trace error representation | |
| Dai et al. | Intelligent audit question answering system based on knowledge graph and semantic similarity | |
| Wehnert et al. | ERST: leveraging topic features for context-aware legal reference linking | |
| Suryachandra et al. | Machine learning approach to classify the sentiment value of natural language processing in Telugu data | |
| CN110619363A (en) | Classification method for subclass names corresponding to long description of material data | |
| Mylavarapu et al. | Assessing context-aware data consistency | |
| CN112434212B (en) | Case-related news topic model construction method and device based on neural autoregressive distribution estimation | |
| Khatun et al. | Analysis of Duplicate Bug Report Detection Techniques |
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 |