TORQUE MODELING IN PERMANENT MAGNET SYNCHRONOUS MOTORS (PMSM) USING ARTIFICIAL NEURAL NETWORKS

Abstract

Applications of artificial intelligence (AI) and machine learning have rapidly expanded, impacting various fields, including electric vehicles (EVs) and related systems such as Vehicle-to-Grid (V2G) and smart grids. These technologies are employed in fault diagnosis, energy consumption forecasting, and optimization of EV charging infrastructure. Models such as artificial neural networks (ANN) and deep learning algorithms, including long short-term memory (LSTM), have been used to enhance the performance of these systems. The application of machine learning is also explored in areas such as operation planning and pricing of charging systems, contributing to a solid research foundation on the subject. In the context of EVs, torque estimation—a challenging metric to obtain—emerges as a key area of interest due to its importance in the performance of permanent magnet synchronous motors (PMSM), commonly used in EVs. This article proposes a feed-forward backpropagation ANN model for torque estimation, utilizing experimental data to assess the model’s performance and accuracy. After training, the developed algorithm achieved a mean error of 2.62% compared to target data upon reaching its 100th epoch. The results obtained highlight the main challenges of implementation and suggest the potential for applying this approach to more complex systems, offering a promising method for optimizing and controlling these systems.