Improved modeling of hydraulic active suspension system in rigid body simulations

Sammanfattning

This thesis aims to improve the simulation model of a hydraulic active suspension systemused in rigid-body vehicle simulations. The study focuses on reducing the mismatch of thesimulation model compared to the real-world damper-station pressure dynamics. The currently implemented first order transfer function is shown to be overly responsive, producingunrealistic output. Four alternative approaches are evaluated: first order models with increased time constants, a tuned second order transfer function including a time delay, asigmoid based model, and a physics-based hydraulic model built in AGX Dynamics. Modelperformance is quantified by comparing model output to measurements via shared-area andovershoot-area metrics in 10% pressure intervals across the terrain dataset. Where realstep-response tests and a terrain driving dataset are used as benchmarks. Increasing thefirst order time constant is shown to not provide a viable improvement. The second orderand sigmoid models reduce some errors but still struggle to reproduce the wanted behavioracross the entire pressure range. The AGX Dynamics model achieves the lowest overshootoverall and better agreement in all pressure ranges making it the most suitable candidate forimplementation. Remaining limitations includes sensitivity at very low pressures and valvecontrol during pressure decreases, motivating future refinement of valve dynamics for betteralignment to the real world suspension system.