This paper demonstrates a method for assessing the optimality of control algorithms for semi-active vehicle suspensions based on multiple performance objectives. A multi-objective genetic algorithm is developed to compare the performance of candidate control algorithms; designers are no longer relegated to relate the performance of control algorithms to simple benchmark controllers or passive responses. Furthermore, this MOGA method provides insights into controller tuning; it can be used to explore the controller gain space and quantify the remaining performance benefits. The approach is applicable to a broad range of dynamic systems, including those where vibration cancelation or attenuation is a performance objective. By modifying the state equations and objective functions, the MOGA can be adapted to find optimal control inputs for systems actuated by smart materials (piezoelectric actuators, shape memory alloys, etc.) that are often utilized in noise and vibration control. In addition, the MOGA is used to investigate a two-dimensional optimal space concerning both ride performance and high-temperature mitigation. Previously studied control algorithms like skyhook and feedback linearization control are shown to approach optimality in one metric, but a control compromise opportunity between the algorithms also remains. Specifically, the skyhook method again displays excellent ride performance, but the feedback linearization control suffers in ride while improving thermal metrics. The MOGA is able to find control inputs that dominate all of the provided algorithms, being closer to the optimal in one or both metrics, without sacrificing either. Unfortunately, the MOGA does not provide a real-time solution for implementing control decisions to move toward optimal. The MOGA's value is evaluating a controller's merits apart from the typical comparisons and revealing the remaining optimality space. Future work will concentrate on finding real-time controllers that are able to operate along the Pareto frontier. Additional work will employ the MOGA on a more expansive set of terrains to provide more robust determinations about control algorithm merits. These future runs may increase the complexity of the model by including multiple degrees of freedom, friction, maximum suspension deflection, and other constraints. While a MOGA provides the Pareto frontier of optimal designs, further work will describe the merits of these optimal combinations and explore other metrics for objective functions. As it is applied to a broader base of real-world evaluations, the MOGA methods and parameters could be refined. This will provide more robust and efficient optimality evaluations.
Multi-objective control optimization for semi-active vehicle suspensions
Multikriterielle Regelungsoptimierung für halbaktive Fahrzeugaufhängungen
Journal of Sound and Vibration ; 330 , 23 ; 5502-5516
2011
15 Seiten, 11 Bilder, 1 Tabelle, 38 Quellen
Article (Journal)
English
Multi-objective control optimization for semi-active vehicle suspensions
Online Contents | 2011
|Control of mechatronic semi-active vehicle suspensions
Tema Archive | 2003
|Mechatronic semi-active vehicle suspensions
Tema Archive | 2002
|Mechatronic semi-active and active vehicle suspensions
Tema Archive | 2004
|Observer-based control of vehicle semi-active suspensions
SAGE Publications | 1999
|