Study On Causing Factors Of Coal Mine Accident Based On Support Vector Machine And Fuzzy-Bayesian Method

China is the largest coal producer and consumer in the world,and coal plays a dominant role in China’s energy structure.Therefore,the safe and stable production of coal is related to the sustainable development of China’s economy,and it is of great significance to ensure the energy security of China.In recent years,China’s coal mine safety production level has made gratifying progress.However,with the deepening of mechanization and informatization of coal mine,coal mine have become a complex nonlinear system with numerous data,deep hidden accident causing factors and fuzzy connection between accidents and causing factors.Therefore,it is very difficult to analyze the causing factors of coal mines and prevent accidents.Therefore,taking the causing factors of coal mine accident as the research object,this paper uses machine learning method to study the causing factors of coal mine accident,which solves the problem of difficult analysis of coal mine accident.And the purpose of this paper is to timely predict the accident existing in the production of coal mine enterprises,effectively investigate the causing factors of accident,and achieve the purpose of preventing and reducing coal mine accident.The research carried out in this paper is as follows:How to quickly and accurately identify the types of coal mine accident is studied.A classification and recommendation model of support vector machine for coal mine accident is proposed to solve problems that the data obtained from the coal mine system is too complex to be analyzed directly through the theoretical model and the accident type is difficult to be predicted accurately.In addition,since the traditional support vector machine is unable to efficiently obtain better model parameters,heuristic algorithm is used to improve it,and the support vector machine parameter improvement method based on particle swarm algorithm and artificial fish swarm algorithm is proposed respectively.After introducing the parameter optimization process of the two algorithms and comparing and verifying the two methods,the artificial fish swarm algorithm support vector machine with better classification recommendation effect was selected for the construction of coal mine accident classification model.Then,the application flow of coal mine accident classification model based on improved support vector machine is presented.How to determine the causing factors’ effects of specific types of coal mine accident is studied.In view of the problems in the coal mine system,such as the strong uncertainty of the dependent variable,the complex formation mechanism of the accident,and the difficulty in the causing factors investigation and analysis,this paper proposes a Fuzzy-Bayesian causing factors analysis model for coal mine accident.Fuzzy theory is applied to Bayesian model to solve the uncertainty of causing factors.By using the method of structural learning and parameter learning,the relation between the causing factors can be better excavated,and the expression and reasoning of the relation between the dependent variables can be more intuitive and clear.Aiming at the defects of traditional structural learning method——K2 algorithm,which is not efficient and easy to learn,this paper improves it by combining with artificial fish swarm algorithm and proposes a structural learning method based on artificial fish swarm-K2 algorithm.In addition,based on the Fuzzy-Bayesian causing factors analysis model of coal mine accident,the calculation method and flow chart of the model are given.The case of X Coal Mine Group is used for verification,and the results show that: The classification and recommendation model of support vector machine for coal mine accident can effectively recommend and identify coal mine accident;The Fuzzy-Bayesian causing factors analysis model of coal mine accident can quickly and accurately determine the occurrence rate of specific types of accident,the key causing factors and the order of screening the causing factors.Both of them will greatly reduce the workload of mine accident prevention,improve the efficiency of causing factors investigation,reduce the occurrence of accident,which have good practical application value.This paper has 38 figures,22 tables and 96 references.