Fixed Switching Frequency Direct Model Predictive Control of Synchronous Reluctance Machines Based on Stator Flux Gradients

Abstract

This paper proposes a direct model predictive control method with a fixed switching frequency for stator flux control of a synchronous reluctance machine. Besides ensuring a fixed switching frequency, the main objective is to minimize the ripple of the stator flux. The objective function, which calculates the switching time instants within the switching period based on the stator flux gradients, is formulated as a standard constrained quadratic programming (QP) problem. Instead of using general-purpose QP solvers, an iterative algorithm based on the active-set method with Lagrange multipliers is proposed to solve this particular QP problem while reducing the computational complexity of the proposed algorithm and enabling successful implementation in real-time embedded systems. The proposed method has been successfully implemented on a laboratory model and the results are compared with a conventional indirect model predictive control. The experimental results show better performance in stator flux ripple and lower total stator current distortion factor for the proposed method compared to the conventional method. A 7.5 kW synchronous reluctance motor drive was used for the experimental validation of the proposed control method.

Keywords

• Constrained quadratic programming problem
• direct model predictive control
• fixed switching frequency
• harmonics
• inverter
• model predictive control with variable switching time instants