A Comparison of Two Methods of Adaptive Nonlinear Model Predictive Control

作     者：A. Bamimore D.A. Akomolafe P.J. Asubiaro A.S. Osunleke 

出 版 物：《IFAC-PapersOnLine》 

年 卷 期：2024年第58卷第25期

页      面：90-95页

主　　题：Adaptive model predictive control linear parameter varying (LPV) model nonlinear predictive control Linear time varying (LTV) model 

摘      要：Conventional nonlinear model predictive control (NMPC) relies on an accurate process model. However, real-world systems’ models are often imperfect due to parametric variations, modelling errors and additive noise, leading to degraded control performance. This study investigates the effectiveness of two adaptive NMPC methods in solving this problem, namely, linear parameter varying (LPV) model predictive control (LPV-MPC) and model predictive control based on successive linearization (SL-MPC). In addition, an innovative approach for identifying LPV models is proposed and applied to three simulation examples. The identified LPV models gave a very strong fit. Also, simulation results demonstrate that both adaptive predictive NMPC (LPV-MPC and SL-MPC) exhibit performance similar to conventional NMPC and superior to Linear MPC. Notably, the two adaptive predictive controllers offer significantly reduced computational time compared to conventional NMPC.