Parameter Estimation of PMSMs Without Signal Injection and Dedicated Tests Considering Voltage Errors and Parameter Variations

This article proposes an offline non-invasive method for the parameter estimation of permanent magnet synchronous motors controlled with non-zero $d$-axis current. The proposed algorithm has been specifically designed to operate in application scenarios where custom signal injection and dedicated test campaigns are not possible and are potentially characterized by scarce measurement data variety. The rank deficiency issue that affects the parameter estimation has been addressed by exploiting data measurements from two different speed/load motor operating conditions (OCs). When more than two OCs are available, the ones used for the parameter estimation are automatically selected to mitigate the estimation errors caused by imperfect inverter non-linearity compensation and parameter variations. The proposed method has been validated and compared with existing non-invasive parameter estimation approaches on experimental datasets collected from two different motors. The results show that the proposed methodology outperforms existing approaches in scenarios characterized by scarce variety of available motor OCs, ensuring moderate estimation errors.