基于粒子群优化卷积神经网络的深基坑变形预测方法

DEFORMATION PREDICTION METHOD OF DEEP FOUNDATION PIT OF CONSTRUCTION ENGINEERING BASED ON PARTICLE SWARM OPTIMIZATION CONVOLUTIONAL NEURAL NETWORK

以华润阜阳中心项目五期总承包项目为研究对象,基于粒子群优化的卷积神经网络法对深基坑围护结构的水平位移和地表沉降进行预测,随着监测时间的增加,深基坑围护结构水平位移量和地表沉降量的预测值与实测值均具有一致的变化规律;与实测值相比,预测围护结构水平位移量的均方根误差为3.89%,平均百分比误差为5.92%,预测地表沉降量的均方根误差为4.53%,平均百分比误差为3.96%,均小于8%的误差限制要求,表明基于粒子群优化的卷积神经网络深基坑变形具有较高的预测精度。

Taking the Phase 5 general contract project of China Resources Fuyang Center Project as the research object,the convolution neural network method based on particle swarm optimization is used to predict the horizontal displacement and surface settlement of deep foundation pit envelope.With the increase of monitoring time,the predicted values of horizontal displacement and surface settlement of deep foundation pit envelope have consistent changes with the measured values.Compared with the measured value,the root-mean-square error of the predicted horizontal displacement of the envelope structure is 3.89%,the average percentage error is 5.92%,and the root-mean-square error of the predicted surface settlement is 4.53%and the average percentage error is 3.96%,both of which are less than the error limit of 8%.The results show that the convolutional neural network based on particle swarm optimization has high prediction accuracy for deep foundation pit deformation.