Abstracts

Virtualiojo jutiklio, skirto transporto priemonės pusiau aktyviai pakabai valdyti, tyrimas: Research on virtual sensor for vehicle semi active suspension control

by Paulius Kojis

The object of this Master’s thesis is the development of a virtual sensor for the vehicle’s unsprung mass vertical velocity. The sensor is based on artificial intelligence and is applied to control the semi-active suspension. Based on the literature review and using MATLAB/Simulink and IPG CarMaker software, a mathematical model of a vehicle with 14 degrees of freedom was developed. Validation of the model was performed using open-access experimental data. The model is equipped with a semi-active suspension and a control algorithm. A data set was generated for virtual sensor learning; it includes 95 scenarios for sensor training, validation and testing. The created sensor uses 14 input parameters to obtain the values of the unsprung masses’ vertical velocities as an output. The robustness and accuracy were evaluated by adding white noise to the inputs of the virtual sensor. The effects of the sensor on the vehicle dynamics were assessed, focusing on the passengers’ comfort. It was found that a person will not sense the difference when replacing a physical sensor with a virtual one. Structure: introduction, three chapters, conclusions, references and appendixes. The thesis consists of 73 pages without appendixes, 37 figures, 10 tables, 61 references and 7 appendixes.