Multibody-based parameter identification of vehicle systems

Details

Supervisors

Bianchin, Gianluca ; Docquier, Nicolas

Faculty

Ecole polytechnique de Louvain

Degree label

Master [120] : ingénieur civil en mathématiques appliquées, à finalité spécialisée

Abstract

enThis thesis addresses the challenge of identifying dynamic parameters in vehicle suspension systems using a multibody modeling approach. Accurate knowledge of parameters such as masses, center of mass positions, and moments of inertia is crucial for high-fidelity vehicle simulations. However, these parameters are often unknown and measuring them directly is extremely difficult due to the complexity of vehicle systems. This work focuses on quarter-car suspension systems, the dynamic behavior of which is modeled using Robotran, a software tool that symbolically derives the dynamic equations of motion for multibody systems. To estimate the parameters, multiple optimization techniques are explored, including the BFGS and trust-region methods. Particular attention is placed on ensuring identifiability through the use of minimal parameter sets and sufficiently rich input trajectories. The identification procedure is validated across multiple quarter-car configurations. The results obtained demonstrate the method’s robustness and accuracy under varying conditions, including measurement noise and limited access to internal system data. This framework offers a practical solution to enhance simulation accuracy in the absence of detailed parameter knowledge.