Controlling Semi-Active Damper for Heavy Duty Trucks: A Comparison of Control Algorithms for Increasing Ride Comfort with Semi-Active Dampers

Sammanfattning

This thesis evaluates semi-active suspension control for a heavy-duty 4x2 truck undervarying payload conditions, with the aim of improving ride comfort while maintainingimplementation feasibility. A rigid full-vehicle model was implemented in IPGTruckMaker and integrated with MATLAB/Simulink. Five damper configurationswere compared: Passive, Load-Adaptive, Skyhook with estimated states, Ideal Skyhook,and a Reinforcement Learning controller based on TD3.The controllers were evaluated on a 1–5 Hz swept-frequency road and on stochasticroad profiles generated according to ISO 8608 classes A–F, using unloaded, 5,000 kg,and 10,000 kg hitch load cases. Performance was assessed primarily through RMS ofvertical acceleration, supported by frequency-domain analysis and power consumption.The results show that the benefit of semi-active control depends strongly on roadroughness. On smooth roads, the passive damper remained highly competitive.On rougher roads, active controllers consistently reduced vertical acceleration relativeto passive damping. Reinforcement Learning achieved the lowest RMS verticalacceleration in many test cases, but required higher power and showed degraded performanceoutside its training distribution. Skyhook with estimated states providedthe best overall balance between comfort improvement, low power consumption, andimplementation feasibility, while Ideal Skyhook offered strong performance at thecost of additional sensing requirements.The Load-Adaptive controller maintained the intended damping ratio across payloadsand was the simplest active strategy to implement at the lowest computationalcost. However, its comfort benefit was limited at high load while power consumptionincreased substantially, suggesting that disabling the load-adaptive logic in operatingregions where passive damping already performs comparably would improve itsenergy efficiency.