Simultaneous Identification of Inverter and Machine Nonlinearities for Self-Commissioning of Electrical Synchronous Machine Drives

Abstract

The proposed identification method allows for a simultaneous estimation of nonlinear output voltage deviations in voltage source inverters (VSIs) and nonlinear synchronous machine models. Based on the identified characteristics with the help of physically inspired structured artificial neural networks (ANNs), an efficient tuning of the current control system can be performed and the nonlinear voltage deviations caused by parasitic effects and dead-time distortions can be accurately compensated for. The identification is performed without position sensor while the rotor is mechanically locked by utilising measured phase currents and reference machine voltages only. Experiments for an interior permanent magnet synchronous machine (IPMSM) and a reluctance synchronous machine (RSM) show that the proposed method is capable of identifying the current dependent self-axis and cross-axis flux linkages, differential inductances and the nonlinear VSI voltage deviations as well as the phase resistance at the same time. The proposed method is fast and generic. Besides the rated machine current, voltage and frequency, no prior system knowledge is required making it applicable for the self-commissioning of any electrical synchronous machine drive.

Christoph Hackl

Professor

My research focuses on the electrical components of renewable and mechatronic energy.