Oil-immersed transformer fault diagnosis method based on multi-source data fusion

The invention belongs to the technical field of transformer fault diagnosis, and relates to an oil-immersed transformer fault diagnosis method based on multi-source data fusion, which comprises the following steps of: selecting five gases as indexes for analyzing gas dissolved in transformer oil, a winding direct-current resistance unbalance coefficient, a winding absorption ratio, a polarization index, an iron core grounding current and a micro-water content; constructing a multi-source information fusion index; the multi-source information fusion indexes and the corresponding transformer fault types form a data set, and the data set is input into an SVM network for training; kernel function parameters and penalty parameters are optimized by adopting a zebra algorithm, and the established ZOA-SVM transformer fault diagnosis model is used for transformer fault diagnosis. According to the method, various state information amounts are analyzed through multi-source information fusion of the transformer, the SVM