Abstract: Based on the analysis of the factors influencing the residual subsidence of a single working face and the available observation data after the maximum subsidence speed, the model for predicting the residence subsidence of abandoned mine goaf is established by adopting L-M BP algorithm. A comparative analysis of the model is made and the results show that it is feasible to predict the residence subsidence of abandoned mine goaf by artificial neural network method, which is of positive significance.

Key words: Abandoned mine goaf, Residual subsidence, Artificial neutral networks