Gas Disaster Prediction Method Research Based On Data Mining And Information Fusion

As an important source of energy, the demand of coal mine resources is increasing in thedomestic and industrial, meanwhile, coal mining has high risk because of its special environment,especially the accident of gas blast which is the top serious accident with the high frequency,widely destructives and bad impact. For a long time, the system has been mainly based on thedetection of gas, which can not be forecast in advance. In this paper, the main problems of thecoal mine gas monitoring are described and discussed, pointing out the importance of andpossibilities to predict disasters. through data mining and analysis with monitoring data, establisha multi-source, multi-platform, multi-sensor coal mine gas disaster online identification modelbased on support vector machine and fuzzy set theory which can do hidden discrimination anddecision-making; it can solve the problem that the gas monitoring system can not realize the gascalamity predicting. The feature level fusion model and algorithm are established with Bayesnetwork as well as D-S evidence theory, in order to resolve the uncertainty and imprecision inthe online identification of the system, at the same time, two models will be test to theeffectiveness. The new theory and method will be proposed which is combined with “datamining”, analysis, processing, fusion and comprehensive judgment of a variety of monitoringinformation. It provides a theoretical basis and scientific methods to solve the predicting problem,which will make a big difference.