Simulation of Ultrasonic Propagation in Transformers within Thermal Fields and Intelligent Methodology for Hot-Spot Temperature Recognition

Abstract

Hot-spot temperature of transformer windings is a crucial indicator of internal defects. However, current methods for measuring the hot-spot temperature of transformers do not apply to those already in operation and suffer from data lag. This study introduces a novel inversion method that combines ultrasonic sensing technology, multiphysics simulation, and the K-nearest neighbors algorithm. Leveraging the penetrative ability and temperature sensitivity of ultrasonic sensing, a detailed physical field simulation model was established. This study extensively investigates the characteristics of ultrasonic wave signals inside transformers. The investigation includes different temperature fields, ranging from 40 °C to 110 °C at 10 °C intervals, and various ultrasonic wave emitter conditions. By extracting the key features of the acoustic signals, such as the peak time, propagation time, and peak amplitude, an accurate inversion of the winding hot-spot temperature is successfully achieved. The results demonstrate that this method achieves a high accuracy rate (98.57%) in inverting the internal winding hot-spot temperatures of transformers, offering an efficient and reliable new approach for measuring winding hot-spot temperatures.