Transformer winding fault diagnosis method based on wireless identification sensing

Abstract

A transformer winding fault diagnosis method based on wireless identification sensing includes the following steps: collecting transformer winding vibration signals in a normal state and when a fault occurs; denoising the transformer winding vibration signals by using singular entropy, and randomly dividing the denoised transformer winding vibration signals in the normal state into two groups, where one group is training data, and the other group is original measurement data; using the denoised transformer winding vibration signals obtained when the fault occurs as original threshold data; reconstructing the original threshold data to obtain reconstructed threshold data, and obtaining a transformer winding residual error threshold when the fault occurs; and reconstructing the original measurement data to obtain reconstructed measurement data, obtaining corresponding residual error data, and comparing the residual error data with the residual error threshold, to implement fault diagnosis on the transformer winding. The present invention implements remote, wireless, and unmanned fault diagnosis on a transformer winding. The diagnosis efficiency is high, the workload of a maintenance person is reduced, and the running efficiency of a transformer is increased.