改进组合模型方法在大坝变形监测数据处理中的应用  

Application of improved combined model method in dam deformation monitoring data processing

作　　者：江波[1] JIANG Bo(Geological Brigade of Guizhou Provincial Geological and Mineral Exploration and Development Bureau,Zunyi,Guizhou 563000,China)

机构地区：[1]贵州省地质矿产勘查开发局一〇六地质大队,贵州遵义563000

出　　处：《测绘标准化》2023年第4期78-83,共6页Standardization of Surveying and Mapping

摘　　要：为了提高大坝变形监测数据噪声抑制及变形预测精度,本文在径向基函数(RBF)神经网络模型的基础上,提出一种利用主成分分析(PCA)优化的PCA-RBF组合变形监测数据处理方法。该方法主要流程为:首先使用PCA方法预处理大坝变形监测数据,通过确定大特征值个数及主分量对监测数据噪声进行抑制,获取有用的变形信号;然后针对RBF神经网络模型中隐含层节点数难以确定的问题,将大特征值个数作为隐含层节点数,提高预测精度。本文利用模拟仿真数据及大坝变形监测数据对本方法的噪声抑制及变形预测性能进行验证,结果表明,本文提出的方法在噪声抑制及变形数据预测性能上比传统的小波变换方法及BP神经网络模型更好,在实际工程应用中具有推广价值。In order to improve the noise reduction and deformation prediction accuracy of dam deformation monitoring data,this paper proposes a PCA RBF combined deformation monitoring data processing method optimized by principal component analysis(PCA)based on the radial basis function(RBF)neural network model.The main process of this method is as follows:Firstly,the PCA method is used to preprocess the dam deformation monitoring data,and noise reduction is performed on the monitoring data by determining the number of large eigenvalues and principal components to obtain useful deformation signals;Then,to solve the problem that the number of hidden layer nodes in RBF neural network model is difficult to determine,the large eigenvalue number is used as the number of hidden layer nodes to improve the prediction accuracy.The noise reduction and deformation prediction performance of the proposed method in this paper are verified through simulation data and dam deformation monitoring data.The results showed that the proposed method has better noise suppression and deformation data prediction performance than traditional wavelet methods and BP neural network models,and has the value of popularization in practical engineering applications.

关 键 词：变形监测 径向基函数 神经网络模型 小波方法 主成分分析 

分 类 号：P258[天文地球—测绘科学与技术]