Active Suspension Using Preview Information and Model Predictive Control

The objective of active suspension is to minimize passenger acceleration, suspension travel, and tire deflections. Although the work done in preview control for active suspension has advanced a great deal over the last 27 years since the pioneering work of Bender (1968), several important issues still need to be addressed. These are: 1. Saturation and rate limiting of actuator control inputs, which are not considered directly in the LQR formulation. 2. Robustness of control design to uncertainties in vehicle dynamics and preview information 3. State estimation and parameter identification to mitigate the effects of sensor noise and parameter variations. 4. Nonlinear dynamics due to friction, hysteresis, backlast, etc. 5. Failure Detection, Identification and Accommodation. 6. Improvements in sensors and actuators for preview information and active control 7. Real-time implementation of control laws on available microprocessors. We propose here for Phase I the methodology of Model Predictive Control (MPC) to address the issues of control saturation, nonlinear dynamics, on-line adaptation and failure accommodation. The remaining issues will be addressed during PHase II. Specific Phase I technical objectives are: 1) Problem formulation, data acquisition and simulation for preview controlled active suspension. 2)Design of robust Model Predictive Control (MPC) 3) Controller tuning, robustness analysis and performance evaluation using simulation and UC Berkeley Active Suspension Test Facility. 4) Implementation and input data requirements for MPC preview controlled active suspension. Prof. Karl Hedrick and his students from the University of California, Berkeley will support the SBIR effort as a subcontractor and provide the use of the active suspension laboratory.