Identification of the Magnetic Model of Permanent-Magnet Synchronous Machines Using DC-Biased Low-Frequency AC Signal Injection

Abstract

This paper proposes a simple procedure for accurate identification of the magnetic model of permanent-magnet synchronous machines through inverter supply. The proposed method accounts for the magnetic saturation and the cross-saturation effects. The identification methods reported in the literature may require a servomotor to drive the motor under test at controlled constant speed or the motor itself must accelerate and decelerate. The technique proposed here can be applied at standstill with or without rotor locking and uses a dc+ac injection strategy to identify the machine inductances to construct its magnetic model. The direct current sets the operating point, whereas the superimposed ac component estimates the inductance at that particular point. Small ac signal is injected to ensure local linearity of the magnetic characteristic. Saturation effects are automatically accounted for by the dc bias level, and cross-saturation effects are quantified through maintaining a constant current along the cross-axis. The magnetic model thus obtained can be used for optimal control of the machine and for accurate torque estimation in vector-controlled drives.