Artificial Neural Network-Based Vector Control for Enhanced Performance of Direct-Drive Wind Energy Conversion System

This paper presents an approach for controlling the machine-side converter (MSC) in a direct-drive wind energy conversion system employing a surface-mounted permanent magnet synchronous generator. The proposed approach utilizes an artificial neural network-based vector control (ANN-VC) to improve the system’s performance under variable wind conditions. The MSC plays a critical role in regulating the generator’s performance, traditionally achieved using PI or PID controllers. However, these conventional methods often struggle with system nonlinearity and varying operating conditions. To address these limitations, the proposed ANN-VC combines the adaptability of neural networks with the robust control framework of vector control. The PI controllers used in the traditional vector control are initially optimized using the Artificial Rabbits Optimization algorithm to ensure optimal performance before comparison with the ANN-based controller. The Levenberg-Marquardt algorithm is used to update the ANN’s weights and biases, ensuring reliable and effective training. The simulation results demonstrate the superior performance of ANN-VC compared to conventional PI-based vector control. This study highlights the potential of ANN-based control strategies to advance the reliability and performance of modern wind energy systems.