Extended Flux Projection Vector Observer Based Position Error Compensation Strategy for Sensorless Direct Torque Controlled PMSM Drives

Accurate estimation of the stator flux is crucial for sensorless permanent magnet synchronous motor (PMSM) drives employing space vector modulation-based direct torque control (SVM-DTC) strategy. However, the voltage-current hybrid model based adaptive flux observer suffers from increased position offset error due to the characteristics of the adaptive law and parameter mismatches. To solve the above problems, a novel compensation strategy based on the extended flux projection vector (EFPV) observer is proposed. On the basis of analyzing the relationship between the mismatch of the hybrid model and the estimated position offset error, the influence of EFPV on the stator flux observation accuracy is revealed. Therefore, a compensation loop constructed by the reduced-order observer for EFPV is proposed to ensure that the position offset error of the adaptive flux observer converges to zero. Furthermore, extended state observer for parameter mismatches is attached to the EFPV observer to enhance the robustness of sensorless drives. The stability of the adaptive flux observer combined with the proposed compensation strategy is analyzed and the proposed strategy is distinguished by the adaptability and universality, as it does not require iterative search optimization or the test signals injection. The effectiveness and feasibility of the proposed method are verified on a 2.2-kW PMSM platform.