Parameter Estimation of Permanent Magnet Synchronous Machines Based on a New Model Considering Discretization Effects of Digital Controllers

Parameter estimation of permanent magnet synchronous motors offers fast adaptation of controllers to changing conditions and thermal monitoring based on tracking resistance estimations. Model-based approaches have been found to be useful toward that goal with a caveat that an accurate estimation can only be achieved with a good model. However, most estimators available in the literature of this kind have been reported to be useful in a restricted operation envelope. In particular, they remain consistent in low-to-moderate speeds and steady-state conditions without compromising acccuracy while only a few expands that limited operation envelope with ad-hoc modifications to simplified models, such as capturing divergence of a simplified model from the reality with a look-up table or adding learning terms to a model. In this paper, based on experimental data from both hardware and simulations, we present a limitation of the existing models that might be the central reason to poor performance of parameter estimators during high speed and dynamic motions. Most importantly, we bring an understanding to that limitation and explain the mechanism behind it analytically. Finally, we propose a new model which is more accurate for improved parameter identification.