Prediction method of deep transfer learning performance of permanent magnet synchronous motor under the condition of few-shot

Abstract

Using deep learning algorithm to establish agent model can realize rapid prediction and analysis of electrical equipment performance, but its training process requires a lot of data. However, the lack of labeled data of modern electrical equipment leads to the low accuracy of the trained deep learning model in predicting the performance of equipment, which restricts the engineering application of the algorithm. This paper establishes a deep transfer learning performance prediction method, which applies the transfer of performance prediction knowledge accumulated in the historical motor data to the performance analysis of the target motor, and completes the performance prediction task of the target motor with less labeled data and computational power consumption.