Intelligent suspension control
Active suspension controller development with AI based Deep Learning algorithms
In the era of autonomous driving, the occupant is the focus of research and development. Because of the reduced involvement of the occupant in driving related procedures, the occupant is more sensitive to any discomfort when the vehicle is moving. As a result, there is a higher demand on improving driver and passenger experience with regard to ride comfort. Active suspensions play a vital role in improving the ride quality because of their ability to exert independent forces on the suspension with the help of separate actuators. The controller development process deals with various challenges such as modelling complexity, physical constraints on the system and real-time implementation.
With significant advances in the fields of Computational Engineering and Artificial Intelligence (AI), control strategies that are more efficient than traditional appraoches are emerging. The AI based methods possess the potential to deal with complex uncertain models and require moderate computational power for real-time implementation. The study focusses on applying Deep Learning approaches to synthesise controllers that learn about improving ride quality by controlling the actuator force. In this process, various analytical vehicle models and Deep Learning algorithms are extensively studied to analyze their potential in terms of performance, robustness and required vehiclular measurements. In the initial phase, these strategies are validated in a simulation environment (Matlab/Simulink, Adams or IPG CarMaker). Subsequently, the performances of the intelligent active suspension controllers are validated on a physical setup such as the quarter car testrig available at the Institute of Automotive Engineering (ika).