Offline Identification of VSI Nonlinearity at Arbitrary Initial Angles based on Inverter Modeling

Identifying and compensating the nonlinearity of voltage source inverter (VSI) has become an important function of high-performance motor drives. However, conventional methods have limitations in applications where the motor is already assembled and rotation is forbidden, because they require rotor pre-positioning before identification to ensure accuracy. This paper proposes an offline identification method for estimating the VSI nonlinearity at arbitrary initial angles. Firstly, a simplified physical (SP) model of the inverter is proposed, based on which the identification errors due to resistance estimation error and zero-axis voltage are revealed. Next, a stepwise identification process based on linear fitting and particle swarm optimization (PSO) is proposed for estimating the parameters of the proposed SP model and resistance. The key lies in two improvements, one of which is simultaneously estimating the SP model parameters and resistance to improve the accuracy of the estimated resistance. The other is improving the identification equation to avoid the estimation error caused by zero-axis voltage. Experiments are carried out on a 3-kW synchronous reluctance motor (SynRM) feed by a VSI to verify the proposed method.