Model Predictive Control of a Multi-Mode Suspension System Using Preview Information and Weight Optimization

Abstract

This study examines the effectiveness of a model predictive controller (MPC) on the multimode active suspension concept to reduce driver absorbed power. A multimode suspension uses additional degrees-of-freedom beyond a traditional active suspension. This allows the suspension to address disturbances with multiple frequencies when coupled with a noncausal control scheme such as MPC. A half-car model is developed by placing both active and passive components in series to mitigate high and low frequency disturbances at the tire-road interface. A stepwise MPC with preview information is applied to the model to measure an incoming disturbance. The model includes noise in state feedback and preview information. A Kalman filter addresses process and measurement noise while the preview information is filtered using an exponential moving average. Process and Measurement noise are considered to be known. Four disturbance profiles are examined: step, two-mode, three-mode, and an ISO8608 Class D profile. Weights used for the MPC process are optimized for each profile according to an objective function based on driver absorbed power. For each profile the response for uncontrolled, controlled, and controlled with preview is examined to assess effectiveness of the multimode suspension. Results demonstrate the utility of the multimode suspension supplemented with MPC for each profile. Driver absorbed power decreases by 68%, 99%, 73%, and 89% for the step, two-mode, three-mode, and ISO8608 Class D profile, respectively. Examination of the natural, disturbance, and forcing frequencies suggests that the MPC accounts for the damped natural frequencies of the system in response to incoming disturbances. The system shows potential to improve performance of vehicles traversing extreme terrain or maintaining a stable chassis for a variety of applications. Future work includes examination of the relationship between the damped natural frequencies, the disturbance, and the resulting control law.