Performance Enhancement of Dual-Star Induction Machines Using Neuro-Fuzzy Control and Multi-Level Inverters: A Comparative Study with PI Controllers

This paper proposes a hybrid speed control strategy for Dual-Star Induction Machines (DSIMs) supplied by Multi-Level Inverters (MLIs). The proposed approach integrates a Neuro-Fuzzy Controller (NFC) with an Indirect Field-Oriented Control (IFOC) technique, leveraging the adaptive learning capabilities of an Artificial Neural Network (ANN) to optimize the NFC parameters. This strategy achieves significant enhancements in speed regulation performance, including a 20% reduction in settling time, a 15% decrease in overshoot, and minimized steady-state error. The NFC's online adaptive learning capability enables real-time adjustments, outperforming the PI controller in handling rotor resistance variations and load disturbances. Simulation results demonstrate a 35% reduction in torque ripple and a 20% improvement in speed regulation compared to PI controllers. The NFC also exhibits faster response times during torque change and remains unaffected by 50% rotor resistance variations. Additionally, the NFC controller achieves up to 51% reduction in Total Harmonic Distortion (THD) compared to the PI controller. Increasing the inverter voltage level from m=2 to m=7 significantly reduces electromagnetic torque ripple, demonstrating a direct correlation between higher inverter levels and improved torque ripple performance. These improvements position the NFC-based strategy as a promising solution for industrial applications requiring precise speed control, such as robotics, electric vehicles, and automation systems.