Prediction of gas concentration based on residual correction of Markov chai

In view of problem of low accuracy of part of prediction values while using gray neural network for gas concentration prediction, the paper proposed a method of using Markov model to correct prediction results of three-order gray neural network model. It described establishment of gray neural network model and Markov residual correction method, and used the method to analyze and predict gas concentration in different locations of a coal mine at different times. Practical application results show that the maximum relative error of predicted gas concentration and measured value reduced from 14% to 6% after Markov residual correction, and the corrected gas concentration curve is closer to the actual changing trend of gas concentration.