基于MATLAB工具箱的基坑深层水平位移神经网络预测

摘要

With practical measurements as sample, prediction model for horizontal displacement of deep foundation pit support structure was constructed by using MATLAB neural network toolbox. The prediction results obtained were in good consistency with the practical monitoring data with less error. Both showed the maximum displacement would be occurred at the depth of 4 m with the same tendency, indicating the constructed model can be applied in the engineering project. The excavation process of deep foundation is affected by multiple factors, such as the hydrogeological conditions, geometry of space, support structure system, construction conditions, construction loads, exposed conditions of foundation and surrounding environment. The influence of each factor on the deformation presents to be a non⁃linear relationship, and it can be seen from the monitored curve that the higher value occurs in the middle and the smaller value occurs at the both ends, indicating the surrounding supporting structure under the action of the axial force, that is in consistency with the on⁃site monitoring result of axial force.%以实测数据为样本，采用MATLAB中神经网络工具箱，建立了深基坑围护结构深层水平位移预测模型。预测结果显示，预测值与实际监测值吻合，误差较小，两者在4m位置处都有最大位移值，趋势一致，表明模型能较好满足工程需要。深基坑开挖过程受水文地质条件、空间几何形状、支撑结构体系、施工条件、施工荷载、基坑暴露状况和周边环境等多因素影响，各因素对变形的影响呈非线性关系，监测曲线不光滑，呈中间大、两头小，反映围护结构受支撑轴力作用，与现场轴力监测结果一致。