Self-Commissioning Algorithm for Inverter Non-Linearity Compensation in Sensorless Induction Motor Drives

In many sensorless field-oriented control schemes for induction motor (IM) drives, flux is estimated by means of measured motor currents and control reference voltages. In most cases, flux estimation is based on the integral of back-electromotive-force (EMF) voltages. Inverter nonlinear errors (dead-time and on-state voltage drops) introduce a distortion in the estimated voltage that reduces the accuracy of the flux estimation, particularly at low speed. In the literature, most of the compensation techniques of such errors require the offline identification of the inverter model and offline postprocessing. This paper presents a simple and accurate method for the identification of inverter parameters at the drive startup. The method is integrated into the control code of the IM drive, and it is based on the information contained in the feedback signal of the flux observer. The procedure applies, more in general, to all those sensorless ac drives where the flux is estimated using the back-EMF integration, not only for IM drives but also for permanent-magnet synchronous motor drives (surface-mounted permanent magnet and interior permanent magnet). A self-commissioning algorithm is presented and tested for the sensorless control of an IM drive, implemented on a fixed-point DSP. The feasibility and effectiveness of the method are demonstrated by experimental results.