Back-Analysis of Initial Geo-Stress Field of an Underground Workshop by BP Neural Network

The accurate determination of initial geo-stress field is the important premise of numerical simulation to the excavation of underground powerhouse. It is usually measured to get the value of initial geo-stress field. But it can not meet the demand of construction by this way. So it is very important to seek the methods that simulates initial geo-stress field according to several measured points. The BP neural network and linear regression are usually used to simulate the initial geo-stress field. In this paper, the combinations of the two methods above are used to solve the problem. The initial geo-stress of an underground powerhouse is back-analyzed by this method. The results show that the results are concordance with the measured data.