Neural-Network Vector Controller for Permanent-Magnet Synchronous Motor Drives: Simulated and Hardware-Validated...

Area of Science:

• Electrical Engineering
• Control Systems
• Artificial Intelligence

Background:

• Permanent-magnet synchronous motors (PMSMs) are crucial in electric vehicles.
• Conventional proportional-integral (PI) vector control suffers from decoupling inaccuracies.
• Advanced control strategies are needed for improved PMSM performance.

Purpose of the Study:

• To develop a neural-network (NN) vector controller for PMSMs.
• To address and overcome the decoupling inaccuracy in conventional vector control.
• To evaluate the NN controller's robustness and adaptability against standard methods.

Main Methods:

• Developed an NN controller using the full dynamic equation of a PMSM.
• Trained the NN using approximate dynamic programming for optimal control.
• Evaluated performance through simulations of electric vehicle PMSMs.
• Validated the controller using an experimental hardware system.

Main Results:

• The NN controller significantly outperformed conventional vector controllers.
• Demonstrated robust and adaptive performance under motor parameter variations.
• Achieved superior results in both simulation and hardware implementations.

Conclusions:

• The developed NN vector controller offers a significant advancement in PMSM current control.
• The NN approach effectively mitigates decoupling issues inherent in PI controllers.
• This technology shows strong potential for enhancing electric vehicle performance and reliability.