Online Detection of Inverter Voltage Error Based on the Voltage Oversampling Measurement and Sigmoidal Function Model

The output voltage distortion of the inverter worsens the performance of the permanent-magnet synchronous motor, therefore, it should be compensated accurately. The inverter output voltage can be measured by the oversampling system, but the measured voltage contains drift and noise. To solve this problem, an adaptive sigmoidal function (S-function) model is first proposed in this article to estimate the voltage distortion. A nonlinear Levenberg–Marquardt regression method with voltage oversampling measurement is employed to identify the initial parameters of the adaptive S-function. During the operation, a recursive least square method is exploited to refresh the parameters of the S-function in realtime. The voltage distortion is accurately obtained by the proposed method, and an accurate compensation scheme is proposed to reduce the current distortion and torque ripple. The effectiveness of the proposed method is validated by experiments.