基于小波阈值降噪的PSO-LSSVM的混凝土坝变形组合模型

Combined Model of Concrete Dam Deformation Based on PSO-LSSVM and Wavelet Threshold Denoising

文章针对已知的PSO-LSSVM监测模型泛化能力低,使得模型预测精度下降的问题,提出了小波阈值降噪的PSO-LSSVM的混凝土坝变形组合模型。此方法是将小波降噪理论和PSO-LSSVM算法结合起来创建的预测精度较高的大坝位移预测组合模型。该模型先是将大坝位移数据进行一遍小波阈值降噪预处理,然后将降噪后的位移数据经过PSO-LSSVM算法进行训练,得到了小波阈值降噪的PSO-LSSVM的混凝土坝变形组合模型。通过某重力坝进行实例验证,证明该监控模型能够准确预测出大坝的位移偏移量,在大坝安全监控方面具有很高的应用价值。

Aiming at the problem that the known PSOLSSVM monitoring model has low generalization ability,which reduces the prediction accuracy of the model, a concrete dam deformation combined model based on PSOLSSVM and wavelet threshold denoising is proposed. This method is a combination of wavelet denoising theory and PSO-LSSVM algorithm to create a dam displacement prediction model with high prediction accuracy. Firstly, the dam displacement data is preprocessed by wavelet threshold denoising, and then the denoised displacement data is trained by PSO-LSSVM algorithm to obtain a concrete dam deformation combined model based on PSO-LSSVM and wavelet threshold denoising. Through the example verification of a gravity dam, it is proved that the monitoring model can accurately predict the displacement offset of the dam, and has high application value in dam safety monitoring.