Enhancing Road Holding and Vehicle Comfort for an Active Suspension System utilizing Model Predictive Control and Deep Learning

Abstract

Active Suspension Systems (ASS) with control are gaining traction as researchers strive for optimal system performance. They are significant in diverse commercial vehicle applications, catering to user demands. This study employs the advanced Model Predictive Control (MPC) technique to enhance the smoothness and safety of a half-car model. The simulation results showed the prowess of MPC controllers under varied control force signal constraints, demonstrating superiority in curtailing vehicle chassis rotation angle and speed by up to 46.93% and 43.34%, respectively. The controller was compared with an artificial neural network controller utilizing only two state signals of the system, trained from MPC data, demonstrating high accuracy with R2 reaching 0.97024 and mean squared error at 7.3557×10-5. This study contributes to the refinement of ASS by focusing on practical implementation and performance enhancement.