Electric drives are at the heart of modern mechatronics, and the industry is rapidly moving toward sensorless control to estimate rotor positions using software rather than physical sensors. This thesis explores how Artificial Intelligence and Neural Networks can revolutionize traditional control architectures. The goal is to investigate and design advanced neural network concepts that learn control and estimation tasks. By mapping system measurements to control states, this research aims to pave the way for the next generation of intelligent, software-defined electric drive control.
- During your thesis you will gain a solid understanding of the physical models of electric drives (e.g., electric machines, inverters) and the simulation environment.
- You will analyze state-of-the-art machine learning and neural network approaches applied to sensorless control of electric drives.
- Furthermore, you will develop novel AI-based control architectures, exploring both modular and end-to-end neural network designs.
- You will implement, test, and validate your control algorithms using high-fidelity electric drive simulation models.
- Finally, you will document your methodology, analyze the results, and present your findings to the development team.