Predication of Gas Emission Based on Wavelet Multi-Resolution Analysis

Gas emission is basically non-stationary time series, Based on this view, wavelet multi-resolution analysis was applied to the predication of gas emission. Firstly, the gas emission data was decomposed by wavelet multi-resolution analysis. Secondly, the single branch reconstruction of each layer was predicted by establishing AR forecasting model. Synthesized all of the results from every layers, the forecasting result was obtained. The simulation showed that predication for mine gas emission with the method of this paper has much higher accuracy than AR model forecasting model.