Research On Human Error Factor Model Of The Fully Mechanized Coal Face Based On Neural Network

Coal as the main energy in China,is closely related with people’s lives.However because of the particularity of coal industry,complex operation environment,and mining staff ideological slack,management personnel negligence,resulting in the frequent occurrence of accidents.Based on the analysis of the major accidents and human factors in China in 1990-2016,it is found that in the past 35 years,the major accidents occurred in coal mines in China are related to human factors.Therefore,it is of great significance to reduce the failure rate of people to reduce the frequency of coal mine accidents and improve the safety and production efficiency of coal mines as the main research object.This paper analyzes the main factors for fully mechanized coal surface failure.Through data analysis,questionnaire and other methods to determine the 23 factors for the failure.The reliability and validity of these factors were tested to ensure the validity and representative of the data.These factors are divided into three levels: the target layer,the middle layer and the factor layer,and the horizontal is divided into internal and external causes of the two parts,which are divided into external factors and environmental factors in the following two aspects.In this paper,introduce the different artificial neural networks models,algorithm and application scope.Finally,the BP neural network is used as the analysis method.The establishment of the three layer(23-11-5)BP neural network model,programming using the neural network toolbox of MATLAB software,selecting 15 representative samples as analysis data,the computer realization of fully mechanized coal face by establishing error model.The validity of the model is verified by testing the Huoerxinhe coal mine.PSO algorithm is used to further improve the BP neural network to form PSO-BP neural network model.By comparing the PSO-BP model and BP model,it is found that the improved model has faster convergence speed,higher precision and better effect.With the continuous filling of sample data,this model will has greater application value,and more practical significance.The research results of this paper are of great significance to reduce human error rate and improve mine safety production.