Identification of VSI Nonlinearity in IoT-Embedded PMSM Drives Using FFT

This paper proposes a novel method to identify the voltage-source inverter (VSI) nonlinearity in internet-of-things (IoT)-embedded permanent magnet synchronous motor (PMSM) drives. The accuracy of existing online approaches for the identification of the VSI distorted voltage is limited by additional harmonic distortion of the voltage references caused by phenomena not considered in the PMSM model. Moreover, methods that rely on the direct compensation of the VSI nonlinearity in the PMSM control scheme cannot be easily implemented in commercial drives since the firmware of the control unit is generally not customizable. The proposed method is designed for contexts in which commercial PMSM drives generate data that are collected and stored in cloud data lakes and custom algorithms cannot be implemented on the drive control unit. To overcome the drawbacks of the existing online approaches, the fast Fourier transform is adopted to compute only the 6-th harmonic components of the d-axis distorted coefficient and d-axis voltage reference corrected using the estimated q-axis stator inductance. The method is validated by using data generated with an accurate Simulink model of a high-speed PMSM drive and comparisons with other methods are provided to show the effectiveness of the proposed approach.