Prediction of Soft Ground Settlement by BP Neural Network

There are many engineering characteristics for soft ground. It is an urgent problem for engineering construction to predict settlement according to the features of soft ground and existing settlement data. In this paper, we used the measured settlement data of soft ground, then come up with the prediction method of soft ground settlement based on the BP artificial neural network in the Matlab simulation environment. This method can modelling highly complex and nonlinear earth structures directly based on real samples. Practical test showed that this method has a high degree of accuracy, and it is very valuable to instruct construction and guarantee the stability of the ground.