WELCOME TO PRECEDE 2023

Tutorial Proposal

“Latest Advances in Robust Predictive Control of Motor Drives”

Instructors:

Dr. S. Alireza Davari, Shahid Rajaee University, davari@sru.ac.ir

Abstract:  

Predictive control has become a flexible and accurate technique in different power electronics applications including the motor drive. Despite various benefits, there are still challenges to using this method in motor drives. An important dispute about model predictive control is the dependency of the method on the load model. When it comes to motor drives application the model includes more uncertainties and more variations. There are two types of uncertainties in motor drives. 1) The parameter mismatch includes any change or variation in the electric or magnetic parameters of the motor or the inverter. For example, stator resistance increases when the motor gets warm. The core saturation is another example. 2) The operating point includes any change in the torque or the speed of the motor. Load disturbance is an example of this category.

There are various cases of research to improve the robustness of predictive control. There are two general methods to reach this goal.

The first technique is based on adaptive model application. This method updates the uncertain parameters during the performance of the system by using adaptive estimation. This method keeps the original model of the motor and inverter and corrects it when the motor is running.

The second approach completely abandons the classic model of the motor and uses a new model called the local model. This technique is known as model-free predictive control in the literature. In this way, the prediction model will not be the classical model of the motor. The local model plays the role of the prediction motor to select the optimum switching state of the inverter. The mathematics of the local model could consist of two concepts. The first kind of the local model is an iterative numerical model. This method uses the previously gathered data to predict future data. The second kind is a disturbance observer. A compact model of the motor in which all uncertainties are lumped into a new parameter as the disturbance. The method applied an observer to estimate the disturbance.

This tutorial will explain different types of robust predictive control in drives applications. A comprehensive comparison will be the conclusion of the tutorial.

Biographies:

S. Alireza Davari received the M.Sc. and Ph.D. degrees both from the Iran University of Science and Technology (IUST), Tehran, Iran, in 2006 and 2012, respectively. Between 2010 and 2011, he left for a sabbatical visit at Technische Universitaet Muenchen, Germany. Between 2013 and 2020, he was at Shahid Rajaee Teacher Training University as an Assistant Professor. Since 2020 he has continued at the same university as an Associate Professor. He was a visiting research professor at Universidad Andres Bello from 2022 to 2023. His research interests include encoder-less drives, predictive control, power electronics, and renewable energy.