基于GEP的桩顶水平位移预测模型研究

Research on the Prediction Model of Horizontal Displacement of Pile Top Deformation Based on Gene Expression Programming

桩顶水平位移的变形监测是一项重要内容,针对传统预测方法存在的不足,本文采用基因表达式编程(GEP),利用Eclipse平台下的Java编程语言,建立了桩顶水平位移预测模型。将灰色GM(1,1)模型、BP神经网络模型和基因表达式编程(GEP)这三种模型对桩顶水平位移的预测结果进行对比分析,得出基因表达式编程不仅能够提高桩顶水平位移的预测精度,而且其学习效率比灰色GM(1,1)模型和BP神经网络模型也有显著提高,从而证明基因表达式编程模型在桩顶水平位移预测方面具有可行性。

Deformation monitoring of horizontal displacement of pile top is an important content. In view of the shortcomings of tradi- tional forecasting methods, Gene expression programming （GEP） is applied in this paper. Using Java programming language of the E- clipse platform. The prediction model of horizontal displacement of pile top is established. The prediction results of GM （ 1, 1 ）, BP neural network and gene expression programming were compared and analyzed. Gene expression programming improves the prediction accuracy and learning efficiency of the horizontal displacement of pile top, compared with the GM （ 1, 1 ） and BP neural network. Therefore, gene expression programming is feasible in the prediction of the horizontal displacement of pile top.