Prediction of goaf settlement with time sequence of wavelet neural network

Abstract

As the rapid development of mining cities and areas, construction land has become so limited with each passing day that abandoned gobs tends to be used inevitably. Monitoring and prediction of ground settlement of mined-out area has significant importance with regard to the development. In this paper, time-series dynamic prediction model of the wavelet network model, improved through artificial neural network (ANN), was applied to study the existing data, which includes 8 prediction factors of horizontal direction x and y and vertical direction h. According to monitoring data of a mine, it is suggested that this model has high fitting precision and stability. The simulating results of prediction agree well with the measurement. Therefore, this model can be adopted to predict the ground settlement, which presents a broaden perspective and is of great theoretical and applied value.