面向大坝变形的重构预测的稀疏表示算法研究

Research on sparse representation algorithm for reconstruction prediction of dam deformation

传感器中采集的大坝原始监测序列不可避免地存在外界及人为造成的噪声,给准确预测大坝变形带来挑战,为解决该问题,提出降噪重构训练集方法来预测大坝变形。针对传统降噪方法受冗余基函数影响,引入K-SVD方法来稀疏表示大坝原始监测序列,自适应地更新原子提高重构的大坝监测序列的有效信息。以真实大坝为例,验证本研究方法的有效性,以预测能力不同的机器学习模型进行试验,试验表明本研究训练集降噪重构算法可以提高大坝变形预测精度,有效展现了大坝变形序列的特征,面对传统降噪算法有着较好的表现,具有一定鲁棒性和可靠性。

The original dam monitoring sequences collected in the sensors inevitably have external and human-caused noise,which brings challenges to the accurate prediction of dam deformation,and to solve this problem,the noise reduction and reconstruction training set method is proposed to predict the dam deformation.Aiming at the traditional noise reduction method affected by redundant basis functions,K-SVD method is introduced to sparsely represent the original dam monitoring sequences,and adaptively update the effective information of the atomically improved reconstructed dam monitoring sequences.Taking the real dam as an example,to verify the effectiveness of this paper’s method,with different prediction ability of the machine learning model for the test,the test shows that this paper’s training set noise reduction and reconstruction algorithm can improve the prediction accuracy of dam deformation,effectively show the characteristics of the dam deformation sequences,in the face of the traditional noise reduction algorithm has a better performance has a certain degree of robustness and reliability.