Skip to main content

Extremum-Seeking-Based Ultra-local Model Predictive Control and Its Application to Electric Motor Speed Regulation

Publication Type
Conference Paper
Journal Name
Publication Date
Page Numbers
56 to 61
Conference Name
2022 IFAC Modeling, Estimation and Control Conference (MECC)
Conference Location
Jersey City, New Jersey, United States of America
Conference Sponsor
Conference Date

Electric vehicle (EV) market is rapidly expanding. As a critical component of EV, an electric motor needs to accurately follow a reference speed signal while respecting the electrical current constraint for safety. Those requirements are usually formulated as a model predictive control (MPC) problem. However, the performance of traditional model-based MPC depends on the accuracy of the system model, which may not always be guaranteed in reality. Therefore, we utilize a data-driven, model-free predictive control strategy, called ultra-local MPC (ULMPC), to control the speed of an electric motor. To further enhance the control performance of ULMPC, we employ the extremum-seeking control (ESC) to tune the control gain of the ULMPC online. Simulation and hardware experiments demonstrate the enhancement of the extremum-seeking-based ULMPC over a constant-gain ULMPC.