基于成分聚类的高阶奇异谱分析及在GNSS监测序列分析中的应用

Higher order singular spectrum analysis based on component cluster and its applications of analysis of GNSS monitoring sequence

高阶奇异谱分析(HSSA)相对于奇异谱分析对不同延时、嵌入维数变化有较好的鲁棒性,在信号处理、工程等领域有着广泛应用。针对高阶奇异谱分析应用中成分序列的选取凭经验过于主观的问题,在理论推导与分析的基础上,提出一种基于成分功率谱聚类的方法,并应用于实际GNSS监测序列的处理与分析,得到了具有明显以一天为周期的周期成分、趋势变化成分以及不规则变动成分,反映了不同因素造成的大桥索塔变形,为后续索塔变形的建模预报、安全评估提供了技术支撑,也说明方法的实用性。

Higher Singular Spectrum Analysis (HSSA) has better robutness than singular spectrum analysis in the different delay ,and embedding dimension changes .HSSA has been used in wide range of applications such as signal processing ,engineering and other fields .For solving the problem of HSSA component sequences selected too subjectively ,a new method is proposed based on component of the power spectral clustering ,which is applied to the processing and analysis of the actual sequence of GNSS monitoring .The result has been obvious to one day for periodic components ,trends and changes in composition of irregular fluctuation component ,reflecting the bridge tower deformation caused by different factors .These works provide technical support for the following tower deformation modeling forecasts ,safety assessment and also approve the applicability of this method .