The Application Study Of BP Neural Network Based On PSO-DE Hybrid Algorithm Of Pcaalgorithm In The Prediction Of Coal And Methane Outburst

Coal is belongs to a widely application of primary energy and fuel in ourcountry, the social economy development and people’s quality of life is closelyrelated to coal in our country. In order to meet the demand of the coal resources,coal mining is increasing very soon, but because the geological condition is verycomplex, plus with the exploitation of technology and safety management is notperfect now, so it will lead to the coal mining become harder, increase the risk ofcoefficient and the growth of the frequency of coal mine accident. The coal andgas outburst is one of the most serious disasters in coal mine accidents, it is athreat to the safety of the underground workers’ life, it restricts the fast andgood development of coal industry. Therefore, the accurate prediction andeffective protection of this highlight accident is a very important work.This paper is carding and analysis the prediction method of domestic andforeign countries, and made a detailed analysis on the reasons of the outburst,it’s rules, it’s conditions, and its important mechanism, all of this provides acertain theoretical basis to how to choose a effective prediction method. For thenonlinear relation between factors which are affecting the characteristics, thetraditional prediction method cannot satisfy the satisfy needs of the mine safetyproduction. In order to solve this problem effectively, on the basis of in-depth study the relevant theories, such as principal component analysis (PCA)、BPneural network, the DE algorithm and PSO algorithm, combining previousresearch results, analyzed their advantages and disadvantages, and mixed the lasttwo algorithm to put forward a new and efficient algorithm, PSO-DE hybridalgorithm. Due to the advantages of the hybrid algorithm can make up for theinadequacy of the BP neural network, so using the algorithm of networkoptimization. BP neural network forecasting model is proposed in this paperbased on the PCA of PSO-DE hybrid algorithm, and the TunLan mine of XiShancoal and electricity as the practical research field, according to the geologicalmining conditions and gas occurrence conditions of the mine, choosing fiveindicators reasonably, through analysis the principal component of theseindicators, to determine two dominant role which have the principal componentscontribution rate (cumulative) not less than85%to replace the5kinds of singleforecasting index, and optimization the two principal components as predictionmodel input values. Through calculation, the predicted results are same to theactually observation, it is proved that this prediction model have goodgeneralization ability and satisfactory accuracy in terms of outburst prediction.According to the forecast results,selecting and implementation a scientific andreasonable prevention and control measures can effectively save the engineeringquantity and the whole process of the outburst prevention, this not only greatlyshorten the cycle of control, improved the outburst prevention effect, but alsolaid a solid foundation for the increase the per unit area yield of coal level,enhance the mine safety management level, improve mine economic benefits.