Fast Response Model Predictive Control for Open-End Winding Induction Motors




Journal Title

Journal ISSN

Volume Title



In automotive testing systems such as chassis dynamometers and engine dynamometers, induction motor is used to provide load torque and to emulate propulsion motors for electrical vehicles. Fast current/torque response and low current/torque ripple are required to precisely evaluate the performance of the vehicle under test.

To reduce the torque ripple, it is necessary to operate the converters at high switching frequencies. A widely used method is to use the multilevel converters. Dynamometers fed by modular multilevel converter (MMC) or neutral point clamped (NPC) converters have been commercialized in industry. However, for MMC case, the individual modules need to be isolated using an input transformer, which is known to be costly and bulky; for NPC case, the capacitors need to be balanced, which results in more complexity in the control. Another alternative is to use the open-end winding topology, which is obtained by disconnecting the neutral of wye-connected induction motor windings, and feeding both sides of the windings by two voltage source inverters (VSI). Due to the interleaved switching of these two inverters, the current ripple can be reduced so that the torque ripple is also reduced. In this topology, since no input transformer is required, the whole system can be cost-effective and of lower volume. And there is no need for capacitor balancing, resulting in simplified control algorithm. However, if the two inverters share the same DC link, zero sequence current is inevitable and needs to be suppressed.

To provide fast torque response, numerous control methods have been proposed in the literature, among which field oriented control using PI controllers (FOC-PI) and direct torque control (DTC) have been successfully commercialized. In FOC-PI, the bandwidth is limited by the PI controllers, and gain scheduling is required if the operating point varies frequently. In DTC, although the structure of controller is simple, the resultant large torque ripple and steady state error limit the use of this method. As conventional control methods have several limitations, model predictive control (MPC) drew increased attention in recent years due to its intuitive concept, fast response, and easy inclusion of system constraints and non-linearities. However, the heavy computation burden and vulnerable parameter sensitivity still remain as problems to be solved in the application of MPC.

This dissertation explores and evaluates the option of using open-end winding induction motor (OEWIM) with model predictive control (MPC) to achieve fast current/torque responses. Four different MPC methods, i.e., linear predictive current control (Linear PCC), non-linear predictive current control (non-linear PCC), linear predictive torque control (linear PTC) and non-linear predictive torque control (non-linear PTC) are proposed. The proposed methods are verified in simulation and experiment. Compared with conventional control methods, the proposed methods achieve fast dynamic responses, better utilization of DC bus, and stronger zero-sequence current suppression.

Aimed at the heavy computation burden in conventional MPC methods, two computational efficient MPC schemes, i.e., predictive current control in A-B-C frame (PCC-ABC) and three-dimensional predictive current trajectory control (3DPCTC), are proposed. The feasibilities of proposed methods are illustrated in simulation and experiment. The results show that the proposed methods reduces the computation time by 61.05% and 64.24% respectively, and achieve stronger zero-sequence current suppression and faster dynamic responses.

To solve the parameter sensitivity issue in conventional MPC methods, a predictive current control with disturbance observer (PCC-DO) is proposed. Compared with conventional MPC methods, the proposed method can effectively response to sudden changes in motor parameter during steady state operation. In the transient tests, the proposed method can accurately compensate the disturbances introduced by stator resistance and magnetization inductance variations, and eliminate the resultant steady state errors.

The proposed OEWIM with MPC approach in this dissertation provides a systematic solution to achieve fast current/torque responses on electrical drives with reduced computation burden and enhanced robustness against parameter mismatches. Additionally, the proposed strategy also increases the cost-effective, fully utilizes the hardware resources, and improves the reliability of the overall system.



Predictive control, Electric motors, Induction—Windings, Torque, Electric vehicles


Copyright ©2017 is held by the author. Digital access to this material is made possible by the Eugene McDermott Library. Further transmission, reproduction or presentation (such as public display or performance) of protected items is prohibited except with permission of the author.