多元宇宙算法在大坝水平位移预测中的应用

Application of multi-verse optimization in horizontal displacement prediction of dam

针对大坝变形预测的影响因素过多且无法精准预测的问题,将主成分分析方法、多元宇宙算法、BP神经网络协同应用于大坝水平位移预测中,构建出PCA-MVO-BP预测模型。通过PCA来降低原始输入参数的维度,消除变量间相关性,并结合多元宇宙算法的快速收敛、泛化能力强的特性,解决BP神经网络预测模型中的权值和阈值的优化问题。以丰满大坝监测数据为测试样本,将组合模型与常规预测模型进行对比。结果表明:相较于常规LSSVM、RF、SVM算法,组合预测模型的平均误差、标准误差、平均绝对百分比误差值均较小,其预测精度较于单一BP神经网络提高了28.85%。表明了PCA-MVO-BP模型在大坝水平位移预测中的现实性。

Aiming at the problem that there are too many influencing factors in dam deformation prediction and cannot be accurately predicted,the principal component analysis method,multi-verse optimization and BP neural network are applied to the horizontal displacement prediction of dams,and a PCA-MVO-BP prediction model is constructed.Through PCA,the dimension of the original input parameters is reduced,the correlation between variables is eliminated,and the weight and threshold optimization problems in the prediction model of BP neural network are solved by combining the rapid convergence and generalization ability of MVO.Using the abundance dam monitoring data as a test sample,the combined model was compared with the conventional predictive model.The results show that compared with the conventional LSSVM,RF and SVM algorithms,the average error,standard error and average absolute percentage error of the combined prediction model are small,and the prediction accuracy is 28.85%higher than that of the single BP neural network.This shows the realism of the PCA-MVO-BP model in the prediction of horizontal displacement of dams.