Estimating Gas Concentration of Coal Mines Based on ISGNN

Online detecting failure of gas sensors in mine wells is an important problem. A key step for solution of the problem is estimating sample values of detected gas sensor, according to sample values of other gas sensors. We propose a scheme based on ISGNN (iteration self-generating neural networks) to estimate gas concentration of coal mines in this paper. First of all, sensors whose correlation information entropy between these sensors and detected gas sensor is the smallest are selected as input features of the classifier. Then ISGNN was employed as a classifier, and estimated sample values of detected gas sensor. Iteration learning self-generating neural network (ISGNN) is an improved self-generating neural network (SGNN). It has inherited the advantages of SGNN, for example, users do not need to set network structures and parameters and it has better precision. Real world gas monitoring dataset was used for experiment. The experimental results show that proposed method is efficient.