Nonlinear and adaptive control algorithms for electrohydraulic active suspension systems

Abstract

This thesis proposes novel control strategies for automobile active suspension systems. The car electrohydraulic active suspension modeling systems were characterized by several phenomena, such as nonlinear dynamics, parametric uncertainties, and uncertain nonlinearities. Therefore, the control laws were developed in this thesis for a SISO quarter car electrohydraulic active suspension, a SISO restricted quarter car electrohydraulic active suspension, and MIMO full car filtered electrohydraulic active suspension. The first control strategy in this thesis focuses on passenger comfort for a quarter car active suspension system. The dynamic system of the quarter car active suspension is known by unmatched model uncertainty. Hence, the control input cannot altogether cancel out the system uncertainties. Thus, a backstepping control system was applied to solve this issue. However, the regression of the backstepping control system is tedious and challenging to determine. So, the Radial basis function neural network system was applied to represent complicated functions. Consequently, an adaptive neural network backstepping control system was developed for SISO semi-strict feedback of a quarter car electrohydraulic active suspension system. The second proposed control strategy addresses the compromise between passenger comfort, road holding, suspension travel limits, and suspension travel oscillations for a quarter car electrohydraulic active suspension. Even though the first control strategy indicated high sprung mass position compensation, it had the worst road holding and suspension travel oscillations. Therefore, we have designed a new model system to explicitly address road holding, which was called a dynamic landing tire system. We have also considered nonsymmetric suspension travel limits instead of that in most previous studies. Accordingly, a nonlinear control filter was developed to track the suspension restrictions. Thus, the second proposed control strategy was the combined nonlinear control filter with the adaptive neural networks control system, which can perfectly deal with the suspension restrictions, dynamic nonlinearities, and system uncertainties. The third control strategy was developed for full car active suspension systems. Although the second control strategy was successfully demonstrated the control effectiveness, it was designed for a quarter car active suspension. Moreover, most previous studies for full car active suspension control systems were not clearly addressed road holding and control robustness. Therefore, the second control system was evolved for the MIMO nonlinear of full car electrohydraulic active suspension system in the presence of both a stiff road perturbation and external aerodynamic disturbances. The third control system consists of the adaptive neural networks backstepping control system and four nonlinear control filters. A zero dynamics system was also applied to guarantee system stability. The simulation results show that the proposed control systems are successfully achieved the control objectives as follows. Firstly, the first control strategy can provide an excellent ride comfort for a quarter car active suspension. Secondly, the second proposed control law can powerfully manage the compromise between passenger comfort, road holding, and suspension travel for a quarter car active suspension. Finally, the third control strategy can achieve the optimal suspension performance of improving passenger comfort, maintaining road holding, avoiding reaching suspension travel limits, reducing suspension travel oscillations, and overcoming dynamic nonlinearities and system uncertainties for full car active suspension.

Date	5 Aug 2021
Original language	American English
Awarding Institution	École de technologie supérieure
Supervisor	Jean-Pierre Kenné (Supervisor) & Honorine Angue Mintsa (Co-supervisor)