Method To Improve a Wheel Suspension Design Using VI-CarRealTime and Reinforcement Learning
Typ
Projektarbete, avancerad nivå
Project Report, advanced level
Project Report, advanced level
Program
Publicerad
2024
Författare
Denneler, Manuel
Heilig, Christoph
Bangalore Venkatesh Prasad , Vinayanand
Madhuravasal Narasimhan, Vivekanandan
Kolekar, Abhishek Amit
Modellbyggare
Tidskriftstitel
ISSN
Volymtitel
Utgivare
Sammanfattning
The project focuses on the enhancement of wheel suspension design through the utilization of VI-CarRealTime and Reinforcement Learning techniques. The primary objective of the study is to improve vehicle dynamics and autonomous systems, thereby contributing to the advancement of automotive engineering. The development of vehicle suspension systems is a complex and iterative process, involving the adjustment of various parameters to meet quantitative and qualitative metrics. The report emphasizes the significance of simulating different suspension setups to achieve optimal design solutions. It highlights the essential collaboration between simulation engineers and design engineers to ensure the successful development of suspension systems.
The project group aimed to use optimisation techniques and artificial intelligence to streamline the process of developing an optimal suspension in a time-saving manner. The use of the VI-CarRealTime simulation tool facilitated the analysis and synthesis loops in the suspension design development process and enabled the evaluation of kinematic properties and system requirements.
Furthermore, this report deals with the application of machine learning theory, in particular with concepts of reinforcement learning. A comprehensive overview of reinforcement learning, its elements, workflows and classification is provided, highlighting its potential for suspension design optimisation. A detailed comparison of reinforcement learning with other optimisation methods is also presented, highlighting its benefits in the context of suspension development.
The development and description of a MATLAB script for the project is presented, highlighting the technical aspects of implementing reinforcement learning techniques in the context of suspension design. This report concludes with a discussion of the potential impact of the research on the automotive industry, emphasising the importance of the results for the advancement of vehicle dynamics and automotive engineering as a whole. To summarise, the project represents a contribution to improving suspension design through the integration of VI-CarRealTime and reinforcement learning techniques. The findings and insights presented in this report have the potential to significantly
impact the automotive industry by contributing to the development of more efficient and optimised vehicle suspension systems.