Forecasting gas emission concentration using wavelet decomposition and GM-ARIMA model.

In order to improve the prediction accuracy of the dynamic gas emission concentration in coal mining and heading working face, an approach based on wavelet decomposition and GM-ARIMA model prediction method was proposed. Firstly, this paper utilized the Grey Model GM (1, 1); the Autoregressive Integrated Moving Average Model ARIMA and the combination prediction model, GM-ARIMA, built by the method of variance reciprocal weighting to forecast the gas emission concentration respectively based on the gas concentration time series, and then employed wavelet theory to decompose the gas concentration time series into approximation sequence and detail sequence, after analyzing the characteristics of decomposition sequences respectively, matching methods-GM (1, 1) and ARIMA model, were employed to forecast the decomposed sub sequences. Then the final prediction values of the gas concentration were obtained through the reconstruction of sub sequences. Finally, it was verified by the engineering application and show that the average error rate of wavelet decomposition and GM-ARIMA prediction method was 3.64%, which provided a better fitting effect and a higher prediction accuracy compared with single model GM (1, 1); ARIMA and GM-ARIMA combination prediction model. It follows that this method is feasible and effective. [ABSTRACT FROM AUTHOR]