基于小波EGM-ISFLA-SVR的大坝变形组合预测模型

为了提高大坝变形监控模型的预测能力,充分挖掘变形实测数据并及时了解大坝的运行性态,提出了一种基于小波EGM-ISFLA-SVR的大坝变形组合预测模型。该模型首先应用小波分析进行去噪,提取变形监测序列的时效分量以及由水压、温度分量组成的综合效应分量。然后,分别运用均值GM(1,1)模型和基于改进的混合蛙跳算法的支持回归机模型对两种序列进行建模和预测。最后,经小波重构得到组合模型。通过工程实例对模型效果加以检验,采用多项指标分别与传统统计模型的拟合精度和预测精度进行对比。结果显示,该模型拟合时具有比统计模型更大的复相关系数和更小的均方差;预测时均方差、平均绝对百分比误差和平均绝对误差均小于统计模型,表明该模型具有更高的拟合和预测精度和更强的泛化能力。

Wavelet EGM-ISFLA-SVR-based hybrid prediction model for dam deformation

In order to enhance the prediction capacity of dam deformation monitoring model，fully mine the measured data of dam deformation and acquire the operation condition of the dam in time，a wavelet EGM-ISFLA-SVM-based hybrid prediction model for dam deformation is proposed herein. In the hybrid model，the wavelet analysis is applied to denoise and then to extract the time dependent component of the deformation monitoring series and the integrated effect component composed by the components of water pressure and temperature at first; afterwards，modelling and prediction are made for both the serieses with the mean value generating GM( 1，1) model and the ISFLA-SVR model respectively; finally，the hybrid model is obtained via the wavelet reconstruction. The effect of the model is verified through the relevant engineering case，in which several indexes are adopted to be compared with the fitting accuracy and prediction accuracy of the conventional statistical model. The result shows that the hybrid model has even larger multi-correlation coefficient and even smaller mean square error than those of the statistical model when fitting，while all the mean square error，the mean absolute percent error and the mean absolute error are smaller than those of the statistical model as well when prediction; which indicates that the model has enven higher fitting and prediction accuracies and even stronger generalization capacity.