Data-driven model predictive control based on discrete space vector modulation for permanent magnet synchronous motor

Area of Science:

• Electrical Engineering
• Control Systems
• Robotics

Background:

• Model Predictive Control (MPC) is crucial for permanent magnet synchronous motors (PMSM) but faces challenges in computational efficiency and parameter mismatch.
• Existing methods often require precise motor parameters, limiting their practical application.

Purpose of the Study:

• To develop an improved MPC algorithm for PMSM that addresses computational burden and parameter robustness.
• To reduce the complexity of voltage vector selection in MPC for PMSM.

Main Methods:

• A novel MPC algorithm utilizing reference voltage vector optimization is proposed.
• Discrete Space Vector Modulation (DSVM) is employed, reducing candidate voltage vectors from 38 to 7.
• A data prediction model based on behavioral system theory is constructed, requiring only input/output data, not exact motor parameters.

Main Results:

• The proposed method significantly reduces the computational load compared to exhaustive DSVM search.
• Experimental results demonstrate enhanced robustness against parameter mismatch.
• The algorithm effectively synthesizes voltage vectors with fewer computations.

Conclusions:

• The developed MPC algorithm offers a computationally efficient and parameter-robust solution for PMSM control.
• This approach is suitable for applications where precise motor parameters are unavailable or subject to variation.