Deep Learning-Based Control of Sensorless Drives for Robotics and Automation

Permanent magnet synchronous machines are critical components in automation and robotics systems due to their compact size, high efficiency, and low maintenance requirements. However, these machines require complex control systems and precise rotor position detection. The signals generated by the rotor position sensor are used to switch the electronic components in the inverter of sensorless drive systems. Although conventional and adaptive controllers are commonly used, their performance can be impacted by changes in system dynamics, necessitating continuous updates to the controller parameters. This research aims to design an intelligent controller based on deep learning, which eliminates the need for a mathematical model of the drive system. The performance of the drive system with the deep learning-based controller will be evaluated and compared with other controllers, such as the adaptive neuro-fuzzy inference system (ANFIS) controller and the traditional PID controller. The obtained results indicate that the intelligent controller based on deep learning outperforms both the conventional and ANFIS-based controllers.