节理岩体地层沉陷的进化神经网络预测方法研究

Prediction Method for Evolutionary Neural Net of Strata Subsidence of Rock Mass

根据神经网络的特点,特别是岩层高度非线性特点,在网络结构的确定和模拟过程中采用了遗传算法,进行全局优化,达到了比较好的效果。

Strata subsidence is one of com-mon geo logic disasters in mines.The high nonlinearity and com-plex of rock strata often make this disaster to be difficultly pre-dicted accurate ly by the existing prediction method.The paper,based on the feature of neural net and the high nonlinearity char-acteristic of rock strata,gives an overall optimization of the neural net with genetic method in the process of the determination and simulation of net structure,and good results were achieved.