The Knowledge of the forces at the contact point between the tyre and the ground allows to maximise vehicle performance, to conduct tyre characterisation and to accomplish fatigue strength analysis on suspension components. In this thesis work, neural networks were implemented to estimate the front and rear wheel loads of a race car from experimental acquisitions. In the first part of the work concerning the front double-wishbone suspension, the performance of the neural networks were analysed in the estimation of the forces at the contact point by receiving, as input, all the strain gauge signals of the suspension arms. Subsequently, the minimum number of instrumented arms were determined, which, integrated with the signals of the inertial platform (IMU) of the car, allowed an accurate estimation of the forces and moments at the wheel. In the second part of the work, the multi-link rear suspension was analysed: using a specifically designed and already constructed test bench, load histories of different circuits were reproduced for the training neural networks to improve the estimation of the wheel forces.
La conoscenza delle forze scambiate al punto di contatto tra pneumatico e terreno permette di massimizzare le prestazioni del veicolo, eseguire la caratterizzazione degli pneumatici e svolgere un’analisi di resistenza a fatica sui componenti della sospensione. In questo lavoro di tesi sono state implementate ed utilizzate le reti neurali per la stima dei carichi alla ruota anteriore e posteriore di una vettura sportiva a partire da acquisizioni sperimentali. Nella prima parte del lavoro riguardante la sospensione anteriore di tipo double-wishbone, sono state analizzate le performance delle reti neurali nella stima delle forze al punto di contatto tra ruota e terreno ricevendo come input i segnali estensimetrici di tutti i braccetti. Successivamente, è stato determinato il numero minimo di braccetti strumentati che, integrati ai segnali della piattaforma inerziale (IMU) di cui è dotata la vettura, permettono la stima accurata delle forze e dei momenti agenti alla ruota. Nella seconda parte del lavoro, è stata analizzata la sospensione posteriore di tipo multi-link: utilizzando un banco prova già progettato e costruito ad hoc, sono state riprodotte storie di carico di differenti circuiti per l’allenamento di reti neurali che permettano di migliorare la stima delle forze a terra.
Applicazione di reti neurali per la stima dei carichi alla ruota di una vettura sportiva a partire da acquisizioni sperimentali
WALOREK, WOJCIECH PIOTR
2021/2022
Abstract
The Knowledge of the forces at the contact point between the tyre and the ground allows to maximise vehicle performance, to conduct tyre characterisation and to accomplish fatigue strength analysis on suspension components. In this thesis work, neural networks were implemented to estimate the front and rear wheel loads of a race car from experimental acquisitions. In the first part of the work concerning the front double-wishbone suspension, the performance of the neural networks were analysed in the estimation of the forces at the contact point by receiving, as input, all the strain gauge signals of the suspension arms. Subsequently, the minimum number of instrumented arms were determined, which, integrated with the signals of the inertial platform (IMU) of the car, allowed an accurate estimation of the forces and moments at the wheel. In the second part of the work, the multi-link rear suspension was analysed: using a specifically designed and already constructed test bench, load histories of different circuits were reproduced for the training neural networks to improve the estimation of the wheel forces.File | Dimensione | Formato | |
---|---|---|---|
Walorek_WojciechPiotr.pdf
accesso riservato
Dimensione
34.52 MB
Formato
Adobe PDF
|
34.52 MB | Adobe PDF |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/36777