Transformer Winding Deformation Detection and Fault Identification Based on Distributed Optical Fiber Sensing

摘要

The traditional detecting methods of winding deformation are off-line detection, and can not identify the winding deformation mode. This paper proposed a new transformer winding deformation detection method based on distributed optical fiber sensing. A fiber-optical composite winding model was designed, the influence of optical fiber on the electric field in oil was simulated, and a transformer winding model with built-in optical fiber was developed. The typical deformations of the transformer were set, using Brillouin optical time domain reflectometry (BOTDR) to measure the strain change of the fiber. Finally, the detection signal was pattern-recognized by the extreme learning machine (ELM). According to the results, the accuracy of ELM was more than 94% for different deformation forms. The distributed optical fiber sensing technology can detect the transformer winding deformation effectively, which provides a new approach for online monitoring of transformer winding deformation.