Integrated control of vehicle chassis systems
This thesis develops a method to integrate several automotive intelligent chassis systems, such as Anti-lock Brake System, Traction Control System, Direct Yaw Control and Active Rear Wheel Steering, using evolutionary approaches. The Integrated Vehicle Control System (IVCS) combines and supervises all controllable systems in the vehicle, optimising the over all performance and minimising the energy consumption. The IVCS is able to improve the driving safety avoiding and preventing critical or unstable situations. Furthermore, if a critical or unstable configuration is reached, the integrated system should be able to recover a stable condition. The control structure proposed in this work has as main characteristics the modularity, extensibility and flexibility, fitting the requirements of a 'plug-and-play' philosophy. The investigation is divided into four steps: Vehicle Modelling, Soft-Computing, Behaviour Based Control, and Integrated Vehicle Control System. Several mathematical vehicle models, which are applied to designing and developing the control systems, are presented. MATLAB, SIMULINK and ADAMS are used as tools to implement and simulate those models. A methodology for learning and optimisation is presented. This methodology is based on Evolutionary Algorithms, integrating the Genetic Leaming Automata, CARLA and Fuzzy Logic System. The Behaviour Based Control is introduced as the main approach to designing the controllers and coordinators. The methodology previously described is used to learn the behaviours and optimise their performance, and the same technique is applied to coordinators. Several comparisons with other controllers are also carried out. From this an Integrated Vehicle Control System is designed, developed and implemented under a virtual environment. A range of manoeuvres is carried out in order to investigate its performance under diverse conditions. The leaming and optimisation method proposed in this thesis shows effective performance being able to learn all the controller and coordinator structures. The proposed approach for IVCS also demonstrates good performance, and is well suited to a 'plug-and-play' philosophy. This research provides a foundation for the implementation of the designed controllers and coordinators in a prototype vehicle.