Deep Learning-Enhanced Direct Torque Control of BLDC Motors for Optimized Electric Vehicle Performance

<div>Electric vehicles (EVs) represent a promising solution to reduce environmental issues and decrease dependency on fossil fuels. The main drawback associated with the direct torque control (DTC) scheme is that it is incapable of improving the efficiency and response time of the EVs. To overcome this problem, integrating deep learning (DL) techniques into DTC offers a valuable solution to enhance the performance of the drive system of EVs. This article introduces three control methods to improve the output for DTC-based BLDC motor drives: a traditional proportional–integral for speed controller (speed PI), a neural network fitting (NNF)-based speed controller (speed NNF), and a custom neural (CN) network-based speed controller (speed CN). The NNF and CN are DL techniques designed to overcome the limitations of conventional PI controllers, such as retaining the percentage overshoot, settling times, and improving the system’s efficiency. The CN controller reduced the torque ripple by 15%, maintained the percentage overshoot by 10–15%, and also improved the settling time by 5%, leading to a 17.5% improvement in energy efficiency compared to the PI controller. The adaptive DL controller provides a 20% faster response time in regulating the torque output during dynamic driving conditions. DL-based DTC speed control improves the BLDC motor performance compared to the traditional PI controllers. The PI controller is simple and efficient for steady-state but shows poor performance in dynamic conditions due to large overshoot and long settling time. The NNF controller improves accuracy in static conditions. The CN controller offers better performance and dynamic flexibility with fast adaptation but requires higher computational power and is more complex to implement. The performance assessments of EVs are validated by developing the FTP72 and US06 driving cycle. This research appears to play a crucial role in advancing propulsion systems for EVs in the future.</div>