Development Of System For Coal And Gas Outburst Forecasting Based On Neural Network

In order to improve the accuracy and reliability of mine and gas outburst forecasting,this paper established the model for forecasting indexes optimizing and the model for outburst forecasting,and developed a forecasting system software of C/S model.Because there are many kinds of forecasting indexes,this paper established a model for optimizing indexes based on Grey Relational Analysis, the most relational indexes are chosen according to the value of Grey Relational Grade,which had important reference to the outburst forecasting model.This paper established two models for outburst forecasting: Fuzzy Clustering model and Neural Network model, weighted the forecasting indexes that had the highest Grey Relational Grade while establishing the Fuzzy Clustering model,and changed the amount of hidden nodes through a simple block of codes under Matlab circumstance to find out the best network structure while establishing the Neural Network model.The forecasting results of outburst ,which came from the two models,was according with the actural status completely. Fuzzy Clustering model could not give a exclusive answer,which was only regarded as reference to outburst forecasting,while the result of Neural Netwok was a number, which was regarded as the final result of outburst danger forecast.SQL Server was linked to VB by ADO,Matlab was connected to VB through ActiveX and m file,Word and Excel were combined to VB utilizing VBA language,in VB circumstance a forecasting system software of C/S model was developed,which consisted of users management,indexes management,samples management,outburst forecasting,outburst prevention and outburst records subsystem.The main function of the system were optimizing forecasting indexes, checking prevention effect and recording outburst,and creating the report of forecasting result automatictly. all of the individual and humanistic design gived a lot of convenience to users,they could weight some indexes in Fuzzy Clustering model,set the number of hidden nodes of BP neural network,modify the default indexes in database freely and conveniently.The system was tested by some indexes and samples coming from some coal mine in HuaiNan,as a result all function of the system achieved designing goals.