Research On Prediction Model Of Coal Spontaneous Combustion Temperature Based On Machine Learning

Coal spontaneous combustion disaster is one of the five major disasters in coal mine.Most mines in China are spontaneous and easy to coal seams.80%?90%of mine fires are aroused byCoal burning.With the development of science and technique,the number of disasters caused by coal spontaneous combustion has decreased year by year,but the prevention and treatment of coal spontaneous combustion disasters is still severe.Therefore,strengthening the research on beforehandadmonition technique of Coal burning and making correct decisions in time for the prevention and cure of Coal burning calamities are of great significance to the security of coalmanufacture and the safety of miners' lives.This article systematically expounds the importance of the study of coal spontaneous combustion early warning methods for the coal industry.And through the large-scale coal spontaneous combustion experiment,coal spontaneous combustion program temperature increase experiment,thermogravimetric experim ent,and finally determined based on field experience CO,O2,CO/AO2,C2H4 and C2H4/C2H6 and other five coal spontaneous combustion early warning indicators.In this paper,the basic analysis of PSO-BP neural network algorithm,support vector machine algorithm,random forest algorithm is carried out,and the applicability of the three algorithms in coal spontaneous combustion temperature prediction is analyzed respectively.In order to test the performance of the coal spontaneous combustion early warning discriminant model,this paper uses the experimental data of coal spontaneous combustion as the learning set and test set of the model.In order to make the learning and prediction of the coal spontaneous combustion early warning discriminant model closer to the true value,this paper carried out the data missing value processing and the data normalization processing on the experimental data,so as to solve the error generated by the coal spontaneous combustion early warning discriminant model during the training process,Make the model achieve higher accuracy.Aiming at the test standard of coal spontaneous combustion discrimination early warning model,this paper introduces four discriminant indicators to test the performance of coal spontaneous combustion early warning discrimination model.Based on the in-depth discussion of the machine learning algorithm,based on the PSO-BP neural network algorithm,support vector machine algorithm,and random forest algorithm,a coal learning spontaneous combustion early warning discrimination model based on machine learning was constructed,and the parameters of the three models were optimized,Matlab was used to compare the performance of the model.The results show that the regression judgment coefficient of coal spontaneous combustion early warning discriminant model based on random forest algorithm reaches 0.8697,which has high accuracy and robustness,and can meet the needs of coal spontaneous combustion.The research results have important research value for early warning and prevention of coal spontaneous combustion.