A methodology for optimizing semi-active suspensions for automotive applications

Electronically controlled suspensions such as active and semi-active systems can both change their suspension characteristics in real time based on varying road situations to improve ride quality, packaging, and road holding. While the performance of active suspensions is superior to that of semi-active suspensions, semi-active suspensions are economical, safe, and require little power.; In this dissertation, we focus on the control law design methodologies of semi-active suspensions. Alternative semi-active controls are proposed, discussed, investigated, and implemented. Since semi-active suspensions do not perform as well as active suspensions due to physical constraints, we attempt to optimize the controller design of a semi-active suspension from various perspectives so as to minimize the effect of the passivity constraints.; The optimal control law in the sense of a quadratic performance index for the deterministic case is first derived. It is then compared, both analytically and quantitatively, with other sub-optimal control laws proposed. Performance index relations among passive, active, and semi-active suspensions are developed for the comparison.; A numerical optimization for stochastic semi-active systems is also studied. Due to the state-dependent time-varying complexity of the system, a controller structure with a linear feedback gain is presumed in the numerical search. The controller performance with the optimal gain determined is then verified in a time-domain simulation.; To further improve the performance of semi-active suspensions, a semi-active suspension with leading system information is investigated. The proposed controller combines the concepts of learning control and predictive control and provides superior performance to that of its leading system. This controller differs from preview controller in that it requires only leading system state/control information but not direct knowledge of disturbances.; The controller proposed above are then implemented on the Berkeley half-car suspension test rig and their performance are verified experimentally. Experimental results show that semi-active suspensions equipped with the proposed controller do provide significant performance improvement over passive suspensions.