Estimation of explosion limits of gas mixture using a single spread GRNN

Abstract

Gas explosion is a very serious hazard. Explosion limits are the important indices to evaluate the safety of multicomponent explosive gas mixture. In order to estimate explosion limits, a single spread generalized regression neural network was employed. The gas mixture consists of six gases, i.e. hydrogen, methane, carbon monoxide, carbon dioxide, nitrogen and oxygen. The number of inputs on the prediction was investigated. The results show that the GRNN model predicted upper explosion limit with good accuracy. However, the prediction of lower explosion limit was poor. The selection of input variables for the GRNN showed significant effect on the predictive accuracy.