Gray wavelet neural network and its application in mining waste prediction

Historical data were recorded in recent 20 years in mining area. But the processing of these small sample data was still a difficult problem. Gray model was considered as a perfect short-term prediction model which had provided a precise output with small sample data, but it was not suitable for and not enough precise for long term prediction. In this paper three models including gray model GM(1,1), evolutional Gray Model (EGM), and wavelet network were introduced and were compared with each other. Then a new neural network named gray wavelet neural network (GWNN) was put forward which combined the advantages of both EGM and wavelet neural network. Morlet wavelet was chosen as the mother wavelet in GWNN building. It was applied in the mine waste output prediction. The results showed that GWNN was suitable for the prediction with the small sample data of fly ash, cinder and coal gangue. Predictions will play an important role in the mining waste management plan in mining area.