Research on PMSM Torque Estimation Based on Improved BP Neural Network Model

Konferenz: MEMAT 2022 - 2nd International Conference on Mechanical Engineering, Intelligent Manufacturing and Automation Technology
07.01.2022 - 09.01.2022 in Guilin, China

Tagungsband: MEMAT 2022

Seiten: 4Sprache: EnglischTyp: PDF

Autoren:
Du, Shuaixiang (Department of mechanical and electrical engineering, Guilin University of Electronic Science and technology, Lingchuan, Guilin, Guangxi, China & Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, Shenzhen, Guangdong, China)
Wei, Shouqi (Department of mechanical and electrical engineering, Guilin University of Electronic Science and technology, Lingchuan, Guilin, Guangxi, China)
Liang, Jianing (Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, Shenzhen, Guangdong, China)

Inhalt:
Accurate torque control is an important technical requirement in joint parts such as manipulators and mechanical dogs. Limited by the occasion, the torque signal is often obtained by the senseless algorithm. In order to solve the problems of low torque estimation accuracy caused by the disturbance of magnet saturation and motor parameter variation in the operation process of permanent magnet synchronous motor, this paper proposes an improved BP neural network model based on the analysis of the mathematical model of the motor and the principle of BP neural network to improve the defects of the BP neural network model. This method embeds the mathematical model into the basic BP neural network model to reduce the scale complexity and training times of the neural network and improve the prediction accuracy. Firstly, the model is built and trained by MATLAB. Secondly, in the training process, the structure of the neural network model is optimized, and the optimal structure of the model is traversed. Finally, the simulation and verification are carried out on the experimental data and platform. The results show that the improved model has less training times and higher accuracy, which improves the prediction performance of the model.