The main objective of this thesis consists in developing intelligent algorithms which can enhance the comfort of high-performance cars. This class of vehicles is equipped with cutting-edge technology, in terms of sensors and actuators, which is not yet fully exploited in its potential and has considerable leeway for scientific research. This thesis focuses on a semi-active suspension system which plays the main role among electronic controllers in shaping the desired vehicle dynamics, and it is responsible for the comfort and handling performance of the vehicle. The driver and the passengers are very sensitive to this system and it is therefore of utmost importance to have a good understanding of its building blocks in order to deliver the best possible performance. First, I present testbench experiments on the damper and a control-oriented model which can describe its behavior. Afterwards, I propose a set of strategies for the control of the vertical dynamics which can significantly improve performance with respect to state-of-the-art algorithms. The proposed logics have been validated on a sports-car; the performance are evaluated subjectively by a professional test driver and objectively by means of the on-board sensing units. Finally, I investigate an automatic calibration framework based on advanced machine learning techniques which can find the optimal calibration of the control parameters directly on the vehicle via iterative experiments; the proposed methodology has been validated on a production vehicle equipped with a proprietary software, and it has been shown to rapidly converge to the optimal calibration within a few experiments.
L'obiettivo di questa tesi consiste nello sviluppare algoritmi intelligenti che possano migliorare il comfort di veicoli prestazionali. Questi veicoli sono equipaggiati con tecnologia all'avanguardia, in termini di sensoristica e attuazione, il cui potenziale non è pienamente sfruttato e hanno quindi un margine significativo per la ricerca scientifica. Questa tesi si occupa del sistema sospensioni semi-attivo, probabilmente il controllore elettronico più influente sulla dinamica veicolo, che è responsabile del comfort e della tenuta di strada del veicolo. Il pilota e i passeggeri sono molto sensibili alla risposta di questo sistema ed è quindi di fondamentale importanza comprenderne i molteplici aspetti per ottenere prestazioni migliori. Inizio presentando le prove a banco sugli ammortizzatori e il modello orientato al controllo che ho sviluppato per descriverne il comportamento. In seguito, propongo una serie di strategie di controllo della dinamica verticale che mi permettono di migliorare significativamente le prestazioni rispetto al corrente stato dell'arte. Le logiche proposte sono validate su un'auto sportiva; le prestazioni sono valutate soggettivamente da un pilota collaudatore e oggettivamente tramite la sensoristica a bordo del veicolo. Infine, ricerco un processo di calibrazione automatica basato su tecniche di machine learning con l'intento di ottenere la calibrazione ottima del controllore elettronico per via di esperimenti ripetuti; questa metodologia è stata validata direttamente su un veicolo equipaggiato con un controllore proprietario attualmente in commercio, dimostrandone la capacità di ottenere una calibrazione ottima in un numero limitato di esperimenti.
Modeling, control and automatic calibration of a semi-active suspension system for high-performance cars
SAVAIA, GIANLUCA
2020/2021
Abstract
The main objective of this thesis consists in developing intelligent algorithms which can enhance the comfort of high-performance cars. This class of vehicles is equipped with cutting-edge technology, in terms of sensors and actuators, which is not yet fully exploited in its potential and has considerable leeway for scientific research. This thesis focuses on a semi-active suspension system which plays the main role among electronic controllers in shaping the desired vehicle dynamics, and it is responsible for the comfort and handling performance of the vehicle. The driver and the passengers are very sensitive to this system and it is therefore of utmost importance to have a good understanding of its building blocks in order to deliver the best possible performance. First, I present testbench experiments on the damper and a control-oriented model which can describe its behavior. Afterwards, I propose a set of strategies for the control of the vertical dynamics which can significantly improve performance with respect to state-of-the-art algorithms. The proposed logics have been validated on a sports-car; the performance are evaluated subjectively by a professional test driver and objectively by means of the on-board sensing units. Finally, I investigate an automatic calibration framework based on advanced machine learning techniques which can find the optimal calibration of the control parameters directly on the vehicle via iterative experiments; the proposed methodology has been validated on a production vehicle equipped with a proprietary software, and it has been shown to rapidly converge to the optimal calibration within a few experiments.| File | Dimensione | Formato | |
|---|---|---|---|
|
20210126_phd_thesis_submission.pdf
non accessibile
Descrizione: Tesi di Dottorato di Gianluca Savaia
Dimensione
12.01 MB
Formato
Adobe PDF
|
12.01 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/177118