基于IWOA-BiLSTM的高拱坝测点群变形预测时空模型OA

In view of the limits of the spatiotemporal model in the deformation prediction of the observation point groups of high arch dam,a novel method for the deformation prediction of observation point group is proposed by combining locally linear embedding(LLE),bidirectional long short-term memory network(BiLSTM),and whale optimization algorithm(WOA).LLE is used for factor dimension reduction,and the input-output relationship of BiLSTM is established.WOA is improved by introducing a nonlinear convergence factor,adaptive weights,Gaussian mutation disturbance,and Tent chaotic disturbance,so as to optimize the hyperparameters of BiLSTM.Subsequently,IWOA-BiLSTM-based deformation prediction model of observation point group is established.A case study is conducted to validate the proposed methodology.The results indicate that:Compared with WOA,the optimization ability,calculation stability,and convergence speed of IWOA are better.The multiple correlation coefficient and residual standard deviation of IWOA-BiLSTM in deformation prediction are 0.989 1 and 0.789 1,respectively,showing the best performance compared with WOA-BiLSTM,BiLSTM,and spatiotemporal model.