Research On Safety Risk Identification And Assessment Of Coal Mine Based On Text Mining

Effective use of modern information technology to strengthen the collection,mining and correlation analysis of safety production big data,improve safety risk identification and assessment capabilities,is of great significance to curb the occurrence of major accidents and ensure production safety needs.The coal industry is a typical traditional energy industry in China.Under the background of the rapid development of information technology,it is a necessary trend for the traditional coal industry to move towards the high-quality development path of the coal industry.Among them,applying modern information technology to solve the major risk identification and assessment problems,promote the intelligentization of coal mine safety production risk prevention and control,and then effectively prevent and resolve major risks of coal mine safety production,and ensure production safety is a very important part.The premise of risk identification and assessment is the acquisition of basic data related to safe production.As a centralized display of accident risk sources,the massive coal mine safety accident investigation reports accumulated over the years are of great value in extracting the key risk factors that may induce coal mine disasters and mining the hidden and unknown risk occurrence rules.If they can be effectively utilized,they can become an important data resource to assist coal mine safety production management.However,there is currently no effective method to deeply mine and utilize the massive unstructured text data of coal mine safety accident cases.Therefore,this paper attempts to introduce a text information mining technology,scientifically integrate risk management theory and modern information technology,and take a large number of coal mine accident case text reports as the analysis object to conduct indepth research on efficient and accurate identification of coal mine safety risk factors,comprehensive assessment of safety risks,and multi-dimensional risk prevention and control strategies,providing a set of data-driven methods that can effectively prevent coal mine safety risks.The main contents and innovative contributions of this thesis are as follows:(1)A risk intelligent identification method for a large number of unstructured coal mine accident case reports is proposed.Starting from the highly non-standardized and unstructured characteristics of text data of coal mine accident cases,the traditional text mining process is optimized.Through the mining steps of Chinese word segmentation,keyword extraction,related word mining,semantic analysis,and accident risk factor component aggregation,the problems in the process of traditional text mining accident risk cause are solved,such as incomplete extraction of key feature information,complex construction of thesaurus,redundant similar expressions.This realizes the efficient and comprehensive analysis and extraction of risk causative information contained in the text data of coal mine accident case reports and the transformation from unstructured accident case data to structured accident risk basic information,which provides method support for the effective use of unstructured safety production data for risk analysis.(2)The association rule mining and Bayesian network method are combined to analyze the correlation and importance of coal mine safety risk factors.Based on the data of coal mine accident cases,the Apriori algorithm is used to mine the highfrequency risk factor sets that affect the occurrence of coal mine accidents and the potential strong correlation among the factors.On this basis,supported by the professional knowledge related to coal mine safety mining,combined with expert knowledge and structural learning,the construction of the coal mine accident Bayesian network model based on the association rules of coal mine safety risk factors is completed.Finally,through the network sensitivity analysis,critical path analysis and statistical frequency analysis,the main risk factors affecting coal mine safety production and their related risk factor sets are obtained,providing decision-making basis for more targeted implementation of coal mine risk factors joint defense control measures to reduce coal mine accidents.The proposed method of constructing Bayesian network structure based on strong association rules between risk factors can help overcome the omission of network nodes and causal relationships between nodes due to expert subjectivity and knowledge limitations,provide a new perspective for datadriven identification of key security risk factors and their complex interaction mechanism research.(3)A coal mine safety risk assessment model based on the combination of t-SNE and ACRO-ELM is established.In view of the structural characteristics of high dimension,large scale and high complexity of accident data,the dimension reduction method of t-SNE high-dimensional data based on nonlinearity is introduced to construct a new coal mine safety risk assessment model combined with Artificial Chemical Reaction Optimization(ACRO)and Extreme Learning Machine(ELM).By analyzing the performance of the model in aspects of prediction effect,error distribution,time cost,etc.,it is found that the t-SNE algorithm is introduced to reduce the dimension of the original data,which can effectively reduce the complexity of the algorithm,solve the problems of overfitting and high training time cost caused by the high complexity of the model,so as to improve the efficiency of mining tasks,improve the learning performance such as evaluation accuracy and evaluation time cost.At the same time,ACRO has more advantages than Genetic Algorithm(GA)in improving the generalization ability and stability of ELM.(4)A specific coal mine safety risk control scheme based on risk assessment is proposed to guide coal mining enterprises to effectively carry out safety production risk pre-control management.Based on the purpose and process of coal mine safety risk control,this thesis discusses how the proposed risk assessment method can be implemented in coal mining enterprises and proposes risk response control measures based on the assessment results.At the same time,based on the identified main risk factors affecting the safety production of coal mines,risk prevention and control measures are proposed from a macro perspective.The proposal of risk control scheme provides a new idea for coal mine safety risk management.