Coalmine Gas Concentration Analysis Based on Support Vector Machine

Abstract

Production safety is concerned continuously in coalmine, especially the gas safety is a key issue in the working management of coal production. With the development of information technology, a large volume of data should collected from sensors deployed in coalmine. Therefore, it is necessary to forecast gas concentration or evaluate the gas safety in the key point, for example, the underground working face, when there are some faults in the sensing system or the data communication system. In this paper, on the one hand, we adopt Support Vector Regression (SVR) to predict gas concentration with the data from other sensors which are running well, on the other hand, we classify the gas concentration data into two class signed to totally safe or a bit high by applying the model constructed by C-Support Vector Classification (SVC) or one-class Support Vector Machine (SVM). Furthermore, Particle Swarm Optimization (PSO) algorithm and Genetic Algorithm (GA) are employed to optimize the parameters of the model. The performances of the models are compared and analyzed in the paper, and the experimental results show that the proposed methods are effective and feasible for processing the gas concentration.