Risk Assessment Of Gas Explosion Disaster Based On Random Forest Model

Coal mine gas explosion is a natural disaster that has the greatest impact on the development of the coal industry in coal mine production accidents,which seriously endangers the life and health and property safety of mine operators.Most coal mine disasters are prevention-oriented,and the risk assessment model for coal mine gas explosion disasters established in this paper lays a theoretical foundation for the accurate pre-control of gas explosions.The specific research results are as follows:(1)Establish a risk evaluation index system.Based on the disaster system theory,the coal mine disaster system is divided into three sub-systems: pregnant environment,hazard factors,and disaster-bearing bodies to study and determine the corresponding secondary indicators of risk assessment.Considering the characteristics of coal mine gas explosion,the gas explosion disaster is finally selected.Three levels of risk assessment indicators.(2)Establish a risk evaluation model.Based on the principle of the random forest algorithm,develop modeling steps,determine risk level demarcation tables,determine prediction errors and importance scoring algorithms,introduce the matlab enhanced toolbox,randomly extract training sets from input samples,train gas explosion classification models,and The model is optimized by a random forest algorithm.(3)Results comparison and case analysis.Import coal mine sample data,use the established risk evaluation model to classify the coal samples of the test set,and compare the results of the gas explosion risk evaluation model based on random forest with the support vector machine model.The accuracy of the random forest model is100%.The error is 0.1308,the accuracy rate of the SVM model result is 75%,and the model error is 0.2156.It is found that the random forest model has smaller errors and higher prediction accuracy,and the random forest model can also quantify the importance of evaluation indicators,including coal spontaneous combustion period Has the highest importance,the average accuracy decreases by 0.1995,the average Gini impurity decreases by 1.514,and the model built is applied to case analysis and targeted comments are given,indicating that the model established in this paper is highly feasible in the risk assessment of coal mine gas explosion disaster.After analysis and demonstration in this paper,it is scientific and reasonable to use coal mines as a disaster system for risk assessment of gas explosion disasters.The risk assessment model of coal mine gas explosion disasters based on random forests is also scientifically feasible,and also provides new ideas for the risk assessment of other disasters in coal mines in the future.