Power consumption behavior analysis method and system based on multi-feature fusion and improved spectral clustering

The invention discloses a power consumption behavior analysis method and system based on multi-feature fusion and improved spectral clustering. The method comprises the following steps: performing data cleaning on power consumption data; extracting power utilization characteristics from the cleaned power utilization data based on a load characteristic curve, signal processing and load characteristic construction; performing feature selection of recursive feature elimination on the power consumption features, and performing feature fusion on the selected power consumption features; based on the feature subset obtained through fusion, using an improved spectral clustering model to classify different power consumption behaviors; the improved spectral clustering model comprising adjacent matrixes which constructs feature subsets based on an enhanced Gaussian kernel function, performing classification according to the adjacent matrixes, according to the enhanced Gaussian kernel function, edge weights being obtaine