High-Performance Switched Reluctance Motor Drive System Using Hysteresis-Based Pulse-Width-Modulation and Cascaded Recurrent Neural Network Controller

The main objectives of the study is to improve speed regulation and reduce torque ripple in Switched Reluctance Motors (SRMs), which are increasingly adopted in modern electric drive systems due to their simple construction, fault tolerance, and wide operating range. Despite these advantages, SRMs suffer from nonlinear magnetization characteristics, acoustic noise, and poor torque performance, especially under dynamic load conditions. These objectives are achieved by developing a robust and intelligent control strategy that integrates a Cascaded Recurrent Neural Network (CRNN) controller with a Hysteresis Current-Controlled (HCC) Pulse Width Modulation (PWM) generator. This hybrid control scheme is supported by a custom-designed (n+1) semiconductor and (n+1) diode power converter topology operating on a 300V DC supply, enabling precise switching and current shaping. The most important results are that the proposed CRNN-based control system exhibits accurate phase current tracking within the hysteresis band and has quick dynamic performance, achieving the reference speed of 2000 rpm in 0.06 s with a rise time of 0.03 s. Under different load circumstances, the steadystate speed error is insignificant. Furthermore, the developed control method significantly decreases torque ripple after 1 second of operation and maintains a smoother torque profile across a wide speed range of 200-2000 rpm, surpassing traditional Proportional Integral (PI) and Fuzzy Logic Controllers (FLCs). The significance of obtained results lies in demonstrating that the proposed neural-networkbased control architecture improves the overall efficiency, reliability, and performance of SRMs, making them highly suitable for high-performance Electric Vehicle (EV) drives and industrial automation systems where precise speed control and minimal torque ripple are essential.