Optimal control strategies for active suspensions in an experimental vehicle

Sammanfattning

This thesis investigates control strategies for active suspensions in KTH’sResearch Concept Vehicle-Dynamic (RCV-D) using the Matlab/Simulinkenvironment.The primary objective is to evaluate various control methods for actuators,aiming to enhance ride comfort by minimizing the vertical motion of thesprung mass, the pitch and roll movements, as well as trying to minimize theenergy consumption of the actuators, making the active suspension systemmore sustainable.Active suspension systems are fundamental in isolating the vehicle’s upperbody from external disturbances, thereby improving ride comfort or enhancingvehicle performance. To achieve realistic simulation outcomes, multiplevehicle models with varying degrees of complexity are examined.Active suspensions make use of actuators to control the motion of the vehicle’scomponents, employing various control strategies for different purposes.Unlike passive suspensions, which consist of non-adjustable springs anddampers that provide a fixed response to external excitations and semi-activesuspensions, which employ adjustable dampers, active suspensions can adaptin real-time. Semi-active suspensions optimize the vehicle response to varyingconditions but their capability is limited since they do not add energy to thesystem. In contrast, active suspensions actively add energy to the system andhave the highest adaptability to external excitations, maximizing the vehicle’sperformance within their physical limits. These systems are highly adaptable,offering superior ride comfort and handling compared to passive and semiactivesuspensions.Various control strategies for active suspensions are analyzed, including HinfinityControl, Proportional - Integral - Derivative (PID) Control, SkyhookControl, Model Predictive Control (MPC), and Reinforcement Learning (RL)-based Control. The aim is to identify the solution that guarantees the best ridecomfort while minimizing energy consumption.