Yu Ruidong ^* Zhang Zhousheng ^*

• Shanghai University of Electric Power, Shanghai, China

The traditional partial discharge localization improvement strategy mainly starts from intelligent algorithms. This paper also considers the influence of the iron core winding and oil temperature. Through finite element simulation, a transformer model was established to analyze the propagation characteristics of ultrasonic signals generated by partial discharge under the interference of gradient oil temperature and winding. It is demonstrated that the ultrasonic velocity generated by partial discharge in transformers cannot be fixed, and it is suggested that ultrasonic sensors should be installed near the center of the top of the transformer. The chaotic firefly-particle swarm hybrid algorithm is proposed, and through the calculation of Shubert's multi-peak function, it is shown that the proposed algorithm can be better optimized in the case of multiple local extreme points. Finally, a partial discharge defect platform based on gradient oil temperature was built to verify the chaotic firefly-particle swarm hybrid localization algorithm. The results indicate that under gradient oil temperature experiments, the algorithm achieves positioning errors less than 100mm and 55mm for cases with and without winding obstruction, respectively, with average positioning errors of 74.2mm and 35.2mm, providing technical reference for practical transformer partial discharge localization during operation.