Offline Parameter Self-Learning Method for General-Purpose PMSM Drives with Estimation Error Compensation

Offline parameter identification of permanent magnet synchronous machines (PMSMs) is of great importance for general-purpose drives with sensorless control. This paper proposes an amplitude-auto-adjusting signal injection (ASI) method for the parameter self-learning of PMSMs at standstill considering inverter nonlinearities and the digital time-delay effect. The ASI method achieves the inductance identification process under various dq-axis currents and at the same time prevents the unexpected rotor rotation during the self-commissioning process. For the test PMSM, the spatial inductance maps of dq-axes and abc-phases concerning the magnetic saturation and cross-coupling effects are identified along with the stator resistance. To enhance the estimation accuracy, an error model of the inverter nonlinearities in dq-axes is established, and a compensation method independent of inverter parameters is proposed based on the Hermite interpolation. In addition, the influence of the digital time-delay effect is analyzed and compensated based on the transient model of the circuits. The effectiveness of the proposed parameter self-learning scheme is confirmed on a 2.2-kW PMSM drive. The accuracy of the experimental results is validated by finite element analysis on the test machine.