Adaptive Neural Network Control Using Nonlinear Information Gain for Permanent Magnet Synchronous Motors

Area of Science:

• Electrical Engineering
• Control Systems
• Robotics

Background:

• Permanent magnet synchronous motors (PMSMs) are crucial in modern applications.
• Existing control methods face challenges with complex dynamics and external disturbances.
• Accurate control and estimation are vital for PMSM performance.

Purpose of the Study:

• To propose an adaptive neural network (NN) control strategy for PMSMs.
• To enhance control and estimation performance using nonlinear information (NI) gain.
• To achieve precise position tracking without the direct-quadrature (DQ) transform.

Main Methods:

• A nonlinear controller designed using a backstepping procedure.
• A three-layer NN approximator to estimate complex functions.
• A novel commutation scheme avoiding the DQ transform.
• NI gains to improve performance under varying load torque and position demands.

Main Results:

• The proposed NN control with NI gain demonstrated improved control and estimation.
• Experimental validation on a PMSM testbed confirmed the method's effectiveness.
• All closed-loop system signals achieved semiglobal uniformly ultimately boundedness (UUB).

Conclusions:

• The adaptive NN control with NI gain offers a robust solution for PMSM control.
• The method effectively handles complex dynamics and external disturbances.
• This approach advances PMSM control technology, particularly for applications requiring high precision.