CN-116861997-B - Logic map analysis method and system based on coal mine case text information

CN116861997BCN 116861997 BCN116861997 BCN 116861997BCN-116861997-B

Abstract

The application provides a logic map analysis method and system based on coal mine case text information. The method comprises the steps of extracting scene events from a pre-obtained accident report sample based on a template extraction algorithm, constructing an initial logic map K 1 of a mine outstanding accident, optimizing accident logic relations in an initial logic map K 1 based on a reinforcement learning algorithm to obtain an application logic map of the mine outstanding accident, and carrying out layout display on map data of the application accident map based on a preset layout algorithm to generate an accident main map of the mine outstanding accident. Therefore, the accident graphic is automatically generated through the accident report sample, and by utilizing the automatically generated accident graphic, professionals can be helped to quickly understand the accident occurrence cause, the reading speed of the accident report is improved, the analysis threshold of non-professionals is reduced, and the efficiency of accident guidance analysis is improved.

Inventors

WANG BO
LI WENHUI
WANG YONGJIE
LI JING
Zhu Jingzheng
WANG BINGSHAN
WANG CHUNXIN
LIU ZHENYU
HUANG YANHAI
ZHANG LEI
Feng Songquan
ZHOU QUANCHAO
MA LONG

Assignees

华能庆阳煤电有限责任公司
华能煤炭技术研究有限公司
中国矿业大学(北京)

Dates

Publication Date: 20260508
Application Date: 20230614

Claims (7)

1. A logic map analysis method based on coal mine case text information is characterized by comprising the following steps: s101, constructing an accident corpus of mine prominent accidents according to an obtained accident report sample, wherein the accident corpus comprises a keyword library and a dictionary library; Extracting key sentences in the accident report sample based on the key word library; Based on the dictionary library, carrying out standardization processing on the key sentences in the extracted coal mine case text to obtain standardization expression of the key sentences of the accident report sample; Inputting standardized expression of the key sentences of the obtained accident report sample into a pre-trained language characterization model, and carrying out vectorization expression on the key sentences; Inputting the vectorization representation of the key sentences as node characteristics into a pre-constructed distinguishing matrix, calculating the similarity between each word in the key sentences forming the normalized expression, so as to perform reduction operation on multiple associated nodes, and obtaining all words meeting a preset similarity threshold in the key sentences forming the normalized expression, wherein the similarity threshold is calculated according to the formula: ; determining the distinguishing matrix Elements of (a) Wherein, the method comprises the steps of, Respectively represent the first of key sentences constituting normalized expression First, second The number of words of the word, Is that And (3) with A kind of electronic device Sentence vector similarity; the multi-association node represents related words in key sentences forming normalized expression; Based on a condensation hierarchical clustering algorithm, clustering all words meeting the preset similarity threshold in key sentences forming normalized expression to obtain a plurality of node clusters of the words meeting the preset similarity threshold in the key sentences forming normalized expression, wherein the plurality of node clusters form an initial logic map ; Step S102, based on reinforcement learning algorithm, the initial logic map is subjected to Optimizing the accident logic relation in the mine to obtain an application logic map of the mine outburst accident; and step 103, carrying out layout display on the map data of the applied logic maps based on a preset layout algorithm to generate an accident main diagram of the mine outstanding accident.
2. The method for logic. The analysis method based on the coal mine case text information according to claim 1, wherein, Based on reinforcement learning algorithm, the initial logic map is subjected to All node clusters and accident logic relations in the map are respectively added, deleted and modified to generate the application logic map 。
3. The coal mine case text information based logic atlas analysis method of claim 2, further comprising: Based on a condensed hierarchical clustering algorithm, logic atlas is applied to the application Clustering operation is carried out to obtain an application logic map of the mine outburst accident ; Based on a dynamic programming algorithm, the application logic map is mapped Is laid out in time order and simulates the application logic graph And obtaining an accident main diagram of the mine protrusion accident according to the accident logic relation among the nodes.
4. The method for analyzing a logic map based on coal mine case text information according to claim 3, wherein in step S103, Based on a condensation type clustering algorithm, according to the application logic atlas Similarity between clusters of nodes of (3), for the application logic map Clustering is carried out again on the node clusters of the mine, the similarity among the clusters after the re-clustering is calculated until the similarity among the clusters is smaller than a preset threshold value of the similarity among the clusters, and clustering operation is finished to obtain an application logic map of the mine outburst accident 。
5. The method for analyzing a logic map based on coal mine case text information according to claim 2, wherein in step S103, Based on a dynamic programming algorithm, the application logic map is mapped The nodes in the map are distributed according to the time sequence, and the map of the application logic is simulated And obtaining an accident main diagram of the mine protrusion accident according to the accident logic relation among the nodes.
6. The coal mine case text information based logic atlas analysis method of claim 1, further comprising: Based on a community discovery algorithm, the logic map is mapped And identifying the node cluster in the mine, and carrying out layout display on the identification result to obtain the accident subdivision illustration of the mine salient accident.
7. Logic map analysis system based on coal mine case text information, which is characterized by comprising: The initial map construction unit is configured to construct an accident corpus of mine prominent accidents according to the acquired accident report samples, wherein the accident corpus comprises a keyword library and a dictionary library; Extracting key sentences in the accident report sample based on the key word library; based on the dictionary library, carrying out standardization processing on the extracted key sentences in the coal mine case text to obtain standardization expression of the key sentences of the accident report sample; Inputting standardized expression of the key sentences of the obtained accident report sample into a pre-trained language characterization model, and carrying out vectorization expression on the key sentences; Inputting the vectorization representation of the key sentences as node characteristics into a pre-constructed distinguishing matrix, calculating the similarity between each word in the key sentences forming the normalized expression, so as to perform reduction operation on multiple associated nodes, and obtaining all words meeting a preset similarity threshold in the key sentences forming the normalized expression, wherein the similarity threshold is calculated according to the formula: ; determining the distinguishing matrix Elements of (a) Wherein, the method comprises the steps of, Respectively represent the first of key sentences constituting normalized expression First, second The number of words of the word, Is that And (3) with A kind of electronic device Sentence vector similarity; the multi-association node represents related words in key sentences forming normalized expression; Based on a condensation hierarchical clustering algorithm, clustering all words meeting the preset similarity threshold in key sentences forming normalized expression to obtain a plurality of node clusters of the words meeting the preset similarity threshold in the key sentences forming normalized expression, wherein the plurality of node clusters form an initial logic map ; An application map construction unit configured to construct the initial logic map based on a reinforcement learning algorithm Optimizing the accident logic relation in the mine to obtain an application logic map of the mine outburst accident; The accident diagram generating unit is configured to perform layout display on the map data of the application logic map based on a preset layout algorithm, and generate an accident main diagram of the mine outstanding accident.

Description

Logic map analysis method and system based on coal mine case text information Technical Field The application relates to the technical field of coal mine safety, in particular to a logic map analysis method and system based on coal mine case text information. Background Coal plays an irreplaceable role in the development of economy and society and plays an important role in an energy consumption system, but mine accidents in the coal mining process cause huge casualties. The method is characterized in that the method is used for researching past accident cases and analyzing the accidents by applying the accident cause theory, the method is a basis for preventing mine outburst accidents, a large number of manual reading modes are adopted in the traditional case cause analysis process, cause information in a description text is identified through past knowledge and experience, the whole working process is boring and uninteresting and takes a long time, meanwhile, analysis results can be influenced by the psychological and physiological fatigue degree, knowledge and experience of readers, and result judgment standards are often inconsistent, so that working errors are caused. Thus, there is a need to provide a solution to the above-mentioned deficiencies of the prior art. Disclosure of Invention The application aims to provide a logic map analysis method and system based on coal mine case text information, so as to solve or alleviate the problems in the prior art. In order to achieve the above object, the present application provides the following technical solutions: The application provides a logic map analysis method based on coal mine case text information, which comprises the steps of S101, performing scene event extraction on a pre-acquired accident report sample based on a template extraction algorithm, constructing an initial logic map K 1 of a mine salient accident, S102, optimizing an accident logic relationship in the initial logic map K 1 based on a reinforcement learning algorithm to obtain an application logic map of the mine salient accident, and S103, performing layout display on map data of the application logic map based on a preset layout algorithm to generate an accident main diagram of the mine salient accident. Preferably, in step S101, the template extraction algorithm is used to extract a scenario event from a pre-obtained accident report sample, which includes constructing an accident corpus of the mine accident according to the obtained accident report sample, wherein the accident corpus includes a keyword library and a dictionary library, extracting key sentences in the accident report sample based on the keyword library, and performing normalized processing on the extracted key sentences in the coal mine case text based on the dictionary library to obtain normalized expression of the key sentences of the accident report sample. Preferably, in step S101, the construction of the initial logic map K 1 of the mine salient accident includes inputting a normalized expression of a key sentence of the obtained accident report sample into a pre-trained language characterization model, vectorizing the key sentence, inputting the vectorized expression of the key sentence as a node feature into a pre-constructed distinction matrix, calculating similarity between each word in the key sentence composing the normalized expression, and performing reduction operation on multiple associated nodes to obtain all words composing the key sentence composing the normalized expression, wherein the multiple associated nodes characterize all words associated with the key sentence composing the normalized expression, and clustering all words satisfying the preset similarity threshold in the key sentence composing the normalized expression based on a condensed hierarchical clustering algorithm to obtain a plurality of node clusters composing the initial logic K 1 of the key sentence composing the normalized expression. Preferably, the formula is as follows: determining an element a * in the distinguishing matrix M *(K1), wherein u i,uj respectively represents the ith and jth words in key sentences forming normalized expression, and Sim BERT(ui,uj) is the similarity of BERT sentence vectors of u i and u j. Preferably, step S102 includes performing operations of adding, deleting and modifying all node clusters and accident logic relations in the initial logic map K 1 based on a reinforcement learning algorithm, so as to generate the application logic map K 2. Preferably, the step S102 further comprises clustering the application logic map K 2 based on a condensation hierarchical clustering algorithm to obtain an application logic map K 3 of the mine salient accident. Preferably, in step S103, based on a condensation clustering algorithm, clustering is performed again on the node clusters of the application logic map K 2 according to the similarity between the node clusters of the application logic map K 2, and th