Aimed at better predicting the deformation of supporting structures of deep foundation pit,the PSO-BP neural network,a BP neural network optimized by particle swarm optimization(PSO)algorithm,was adopted to optimize the selection of the weights and thresholds in the normal BP neural network.Therefore,the PSO-BP could avoid some disadvantages of BP neural network,such as being easy to fall into local extreme point and slow convergence.Then,based on an actual project,the optimized and unoptimized BP neural network have been used to predict the horizontal deformation of deep foundation pit wall.In the simulation,the complex nonlinear system equation of monitoring stress and horizontal deformation of foundation pit wall are designed as a black box in the model.The comparison of these two methods show that,the PSO-BP can obtain better prediction of the deformation value of the sup-porting structures,which could be a better tool for future practical engineering.
关键词
深基坑/变形预测/BP神经网络/粒子群优化算法
Key words
deep foundation pit/deformation prediction/BP neural network/particle swarm optimization
维普
万方数据