高阶奇异谱分析在变形监测数据中的应用

Application of High Order Singular Spectrum Analysis inDeformation Monitoring Data

如何从监测得到的GNSS时间序列中获取有用信息,了解变形监测对象的变形特征,已经成为变形监测领域一项重要研究课题。为了提取桥梁GNSS监测时间序列中桥梁变形特征,针对奇异谱分析在鲁棒性上的不足,本文提出了利用鲁棒性更好的高阶奇异对特长钢箱梁桥GNSS监测数据进行处理。通过两种评价指标对奇异谱分析与高阶奇异谱分析的信息提取效果进行量化对比,结果表明利用高阶奇异谱处理后得到的时间序列更平滑,与原始时间序列的相关程度越高,去噪效果越好。此外对经高阶奇异谱分析处理过滤的噪声进行标准正态分布检验,结果表明经高阶奇异谱分析过滤掉的噪声呈标准正态分布,桥面属于稳定状态。

How to obtain useful information from GNSS time series obtained from monitoring and understand deformation characteristics of deformation monitored objects has become an important research topic in deformation monitoring field.In order to extract bridge de-formation characteristics from GNSS monitoring time series of bridges,and in view of the lack of robustness of singular spectrum analy-sis,this paper proposes to process GNSS monitoring data of long-length steel box girder bridges by using higher-order singularities with better robustness.Quantitative comparison is carried out by two evaluation indexes of information extracted by singular spectrum analysis and high-order singular spectrum analysis.The results show that the time series processed by the higher order singular spec-trum is smoother,the correlation degree with the original time series is higher,and the denoising effect is better.In addition,the standard normal distribution test is carried out on the noise filtered by the high-order singular spectrum analysis.The results show that the noise filtered by the high-order singular spectrum analysis is the standard normal distribution and the bridge deck is stable.