基于小波去噪的变形监测预测模型优化研究  

Research on the Optimization of Deformation Monitoring and Prediction Model Based on Wavelet Denoising

作　　者：王水清 张明栋 张明智 WANG Shuiqing;ZHANG Mingdong;ZHANG Mingzhi(Shenzhen Aihua Survey Engineering Co.,Ltd.,Shenzhen 518000,China;Shenzhen Institute of Surveying and Mapping(Group)Co.,Ltd.,Shenzhen 518000,China)

机构地区：[1]深圳市爱华勘测工程有限公司,广东深圳518000 [2]深圳市勘察测绘院(集团)有限公司,广东深圳518000

出　　处：《测绘与空间地理信息》2025年第2期198-201,204,共5页Geomatics & Spatial Information Technology

摘　　要：采用小波阈值去噪方法对原始变形监测数据进行异常数据剔除插补,分别以去噪前后的数据序列构建灰色GM(1,1)模型和BP神经网络模型,并对去噪前后模型预测结果进行对比分析,结果表明:小波去噪后灰色GM(1,1)模型精度大大提升,预测结果变形特征与变形其实与实际观测值更为吻合;小波去噪前BP神经网络模型预测结果较为准确,但小波去噪后模型精度依旧有一定程度的提升。因此,采用小波去噪方法剔除原始观测数据中的噪声影响,能够有效提升变形监测预测模型的精度,提高预测结果的准确性。In this paper,the wavelet threshold denoising method is used to remove and interpolate abnormal data from the original deformation monitoring data.The grey GM(1,1)model and BP neural network model are constructed based on the data sequences before and after denoising,and the model prediction results are compared and analyzed.The results show that:After wavelet denoising,the accuracy of the grey GM(1,1)model is greatly improved,and the deformation characteristics and deformation of the predicted results are more consistent with the actual observed values.The prediction results of BP neural network model before wavelet denoising are more accurate,but the accuracy of the model is still improved to a certain extent after wavelet denoising.Therefore,the wavelet denoising method can effectively improve the accuracy of deformation monitoring and prediction model and improve the accuracy of prediction results by eliminating the influence of noise in the original observation data.

关 键 词：小波去噪 灰色GM(1 1)模型 BP神经网络模型 精度分析 

分 类 号：P25[天文地球—测绘科学与技术] TB258[一般工业技术—工程设计测绘]