基于LMD-PSO-LSSVM组合模型的深基坑变形预测

变形是造成基坑安全隐患的重要因素。为准确预测基坑变形趋势,提出一种将局部均值分解(LMD)、粒子群优化算法(PSO)与最小二乘支持向量机(LSSVM)组合的深基坑变形预测模型。通过 LMD 将时序样本分解为多个分量,利用PSO优化后的LSSVM模型对各分量建立非线性基坑变形预测模型,最后采用滚动预测的方法对各分量进行预测并将结果叠加得到时序样本的预测值。通过实际工程进行模型预测与分析。结果表明:该模型不仅反映出基坑变形本质特征,而且预测精度明显提高,将其运用于基坑变形预测研究中具有较好的应用性和可靠性。

Deep Foundation Pit Deformation Prediction Based on LMD-PSO-LSSVM Model

Deformation is an important factor of foundation pit safety hidden trouble. To predict the deformation trend, a local mean decomposition (LMD), particle swarm optimization algorithm (PSO) and least squares support vector machine (LSSVM) combination forecasting model of deep foundation pit deformation is proposed. Firstly, time-series samples was decomposed into multiple components by LMD. Secondly, PSO was used to optimize the LSSVM model to build the components in the nonlinear deformation prediction model. Finally, the method of rolling forecasts was adpoted to predict for each component, and after superposing sequential sample predictive value was obtained. Through the actual engineering model prediction and analysis, the results show that the model not only reflects the essential feature of deformation, but also obviously improves the prediction accuracy, and applying it in foundation pit deformation prediction research has good applicability and reliability.