极限学习机模型在季冻区深基坑地表沉降预测中的应用

针对传统数据驱动模型存在收敛速度慢、过度拟合等问题,提出了基于极限学习机算法的基坑地表沉降预测方法。结合季冻区地铁车站基坑的特点,提取基坑开挖时间、开挖深度、围护桩顶位移、围护桩内力、支撑轴力及地表温度等特征信息,建立极限学习机回归预测模型,选用实例数据进行算例分析,并将其与传统回归预测模型进行对比,实验结果表明,极限学习机模型收敛速度快,泛化能力强,其预测精度优于传统预测模型,且在学习速度方面优势明显,对深基坑安全监控有一定的实用价值。

Application of extreme learning machine model in ground settlement prediction of deep foundation pit in seasonal frozen area

In order to solve the problems of slow convergence rate and over fitting in driving model using traditional data, a prediction method of ground settlement of foundation pit based on extreme learning machine algorithm was proposed. Combined with the characteristics of subway station foundation pit in seasonal frozen area, the regression prediction model of extreme learning machine was established by extracting the characteristic information of excavation time, excavation depth, top displacement of retaining pile, internal force of retaining pile, supporting axial force and surface temperature, etc. The prediction model was analyzed by instance data, and compared with the traditional regression model. The results showed that the extreme learning machine method converged faster, and had higher generalization ability and learning speed. Moreover, the prediction accuracy was better than the traditional prediction model. Therefore, the model can play a significative role on safety monitoring of deep foundation pit.