Application of Least Squares Support Vector Regression Based on Time Series in Prediction of Gas

Abstract

Prediction of gas emission has received much attention on the security of production of coal mining during the last decade. Gas emission is a random time series during the coal mining and the abnormal values can cause some dangerous events. We analyze these time series data and develop an online LS-SVR algorithm named TOLS-SVR based on time series data. Time series data from gas emission are converted into a set of vectors using theory of time series. The optimal embedding dimension is determined through cross validation technique. Experimental results demonstrate the good performance and efficiency of the proposed TOLS-SVR compared with the traditional online LS-SVR for gas emission prediction.