Intelligent Prediction Method of Hot Spot Temperature in Transformer by Using CNN-LSTM&GRU Network

Abstract

In this paper, an intelligent prediction method of hot spot temperature in transformer abnormal thermal diffusion by using CNN-LSTM&GRU network is proposed. With the continuous development of power grid, as an important equipment in transmission line, the stable operation of transformer is very important. However, the increase of power load demand leads to frequent transformer accidents in recent years, among which, hot spot temperature is the key factor causing transformer thermal aging and even fire. Due to the anisotropy of transformer materials, the thermal diffusion of transformer is an abnormal diffusion process, making the traditional method difficult to predict the hotspot temperature efficiently and accurately. Therefore, this paper studies a deep learning algorithm based on CNN-LSTM&GRU network to predict transformer hot spot temperature. We conduct experiments, and the final results indicated the performance of our model is better than that of the traditional approach in transformer hot spot temperature prediction tasks.