A spatiotemporal identification method for deformation characteristics of expansive soil canal slope based on spectral clustering

Structural health diagnosis of expansive soil slopes requires timely analysis of deformation monitoring data. A method for spatiotemporal clustering of monitoring data for health diagnosis is proposed. First, the deformation time series is upgraded to a panel time series, which includes spatial positions and temporal variations, and similarity characteristics of spatiotemporal deformation are discussed. Second, a similarity distance indicator is defined using three deformation variables: weighted absolute distance, weighted increment distance, and weighted growth rate distance. Third, a spatiotemporal clustering model of the deformation of expansive soil slope based on a spectral clustering algorithm is developed, together with a scoring algorithm for determining optimal clusters. The method analyses and diagnoses the deformation behaviour of the expansive soil slope structure of China's South-to-North Water Diversion Project central line. The advantage of the proposed method is demonstrated by comparing its results with results obtained by the commonly used temporal clustering method. It is further shown how the new method can be used to identify abnormal regions of expansive soil slope deformation.