GPU based Stochastic Parameterized NMPC scheme for Control of Semi-Active Suspension System for Half Car Vehicle ?

摘要

Control of complex systems with inherent randomness in process dynamics poses a serious concern for control engineers, especially in situations where performance and constraint satisfaction are highly demanded. In this paper, we propose a real time (RT) scenario based stochastic parameterized NMPC (SS-pNMPC) scheme for control of semi-active (SA) system for a half car vehicle. The method utilizes graphic processing unit (GPU) to generate several RT scenarios of the random road profile for each parameterized input and through Monte-Carlo (MC) simulations, the expected objective function along with a probabilistic constraint violation certificate are numerically obtained. The optimal input is elicited by finding the input either with minimum expected objective or with the lowest probabilistic constraint violation certificate. The method was implemented on NVIDIA Jetson embedded boards and also, tested in MATLAB/Simulink environment for different ISO road profiles and the simulation results exhibits better performance of the proposed method in comparison to passive systems.