An Accurate VSI Nonlinearity Self-Learning Method of Dual Three-Phase PMSM Drive Based on Voltage Error Transformation Between VSD Subspaces

Voltage source inverter (VSI) nonlinearity self-learning has become an important function of high-performance motor drives, which is conducive to sensorless control and parameter identification. In this paper, an accurate VSI nonlinearity self-learning method based on error voltage transformation is proposed for dual three-phase permanent magnet synchronous motor (DTP-PMSM) drives. First, the characteristics of phase voltage error and zero-axis voltage are analyzed, based on which the voltage error transformation between $xy-\mathbf{plane}$ current injection $(xy-\mathbf{PCI})$ and $dq-\mathbf{plane}$ current injection $(dq-\mathbf{PCI})$ is proposed. Then the self-learning current injection sequence in the $xy$ plane is designed, and the voltage errors of $xy-\mathbf{PCI}$ are transformed into that of $dp$ by the proposed transformation. Finally, the 2D linear interpolation is adopted to compensate for the inverter nonlinearity under different rotor positions and $dq- \mathbf{plane}$ current. The proposed method well addresses the problem of zero-axis voltage and has higher accuracy compared with the existing method. Experimental validation is conducted to verify the effectiveness of the proposed method.