Noise is the second most significant cause of pollution inWestern Europe and it can cause several health diseases. Among the main sources, tyre/road noise is the predominant component of this environmental problem. To limit it, strict restrictions have been introduced and tyre manufacturers have been forced to invest several resources in the study of solutions to reduce their products’ noise and meet the legislative requirements. In such a context, this thesis, as part of the AirBorne Exterior Noise (ABEN) project between Politecnico di Milano and Pirelli Tyres, is aimed at developing predictive tools to investigate and estimate the noise of different tyre prototypes during their design stage. In particular, the goal is an accurate estimation of the tyre noise level in every one-third octave band within 20 - 4000 Hz and the evaluation of three key performance indexes. For this purpose, a black-box modelling approach through Artificial Neural Networks is adopted. The complete work is divided into stages to achieve the goal. At first, an experimental champaign is performed, and the parameters correlated to tyre noise are identified and extracted from the measurements, resulting in a database of 200 tyres. Meanwhile, specific pre-processing techniques are applied to extract the meaningful features from the tread pattern drawing of each tyre in the database. Secondly, the artificial neural network is defined and trained based on the database, and improved according to its predicting performance. Eventually, the obtained outputs are postprocessed to compute performance indexes and evaluate the accuracy of the results.
Il rumore è la seconda causa di inquinamento nell’Europa occidentale e può essere causa di diverse malattie. Nonostante ci siano più sorgenti, la principale componente di questo problema ambientale è il rumore emesso dal rotolamento dello pneumatico sull’asfalto. Per limitare ciò, sono state introdotte delle rigide restrizioni ed i produttori di pneumatici sono stati costretti ad investire molte risorse nello studio di soluzioni per ridurre il rumore causato dai loro prodotti e per rispettare i requisiti legislativi. In questo contesto, il presente lavoro di tesi è parte del progetto AirBorne Exterior Noise (ABEN) del Politecnico di Milano in collaborazione con Pirelli. L’obiettivo di questo progetto è lo sviluppo di strumenti predittivi per studiare e stimare il rumore dei prototipi di pneumatici durante la fase di progettazione. In particolare, l’obiettivo finale di questo studio riguarda la stima accurata del livello di pressione sonora data dal rumore dello pneumatico in terzi d’ottava nell’intervallo di frequenza tra i 20 Hz e i 4000 Hz e la valutazione degli indici chiave di prestazione. Per raggiungere questo scopo viene adottato l’approccio del modello a scatola nera, attraverso l’uso di reti neurali artificiali. Ci sono diverse fasi necessarie per arrivare all’obiettivo. In un primo momento è stata eseguita una campagna sperimentale e sono stati individuati ed estratti dalle misurazioni i parametri correlati al rumore dello pneumatico, risultando in un banca dati costituita da 200 pneumatici. Inoltre, in fase di pre-elaborazione, sono state applicate specifiche tecniche per estrarre caratteristiche significative del disegno battistrada di ogni pneumatico presente nella banca dati. Successivamente la rete neurale artificiale è stata definita e allenata. Infine, i risultati ottenuti sono stati post-elaborati per calcolare gli indici di prestazione e valutarne l’accuratezza.
Predictive modelling of tyre rolling noise through an artificial neural network
Dinosio, Arianna
2020/2021
Abstract
Noise is the second most significant cause of pollution inWestern Europe and it can cause several health diseases. Among the main sources, tyre/road noise is the predominant component of this environmental problem. To limit it, strict restrictions have been introduced and tyre manufacturers have been forced to invest several resources in the study of solutions to reduce their products’ noise and meet the legislative requirements. In such a context, this thesis, as part of the AirBorne Exterior Noise (ABEN) project between Politecnico di Milano and Pirelli Tyres, is aimed at developing predictive tools to investigate and estimate the noise of different tyre prototypes during their design stage. In particular, the goal is an accurate estimation of the tyre noise level in every one-third octave band within 20 - 4000 Hz and the evaluation of three key performance indexes. For this purpose, a black-box modelling approach through Artificial Neural Networks is adopted. The complete work is divided into stages to achieve the goal. At first, an experimental champaign is performed, and the parameters correlated to tyre noise are identified and extracted from the measurements, resulting in a database of 200 tyres. Meanwhile, specific pre-processing techniques are applied to extract the meaningful features from the tread pattern drawing of each tyre in the database. Secondly, the artificial neural network is defined and trained based on the database, and improved according to its predicting performance. Eventually, the obtained outputs are postprocessed to compute performance indexes and evaluate the accuracy of the results.| File | Dimensione | Formato | |
|---|---|---|---|
|
2021_10_Dinosio.pdf
non accessibile
Dimensione
8.76 MB
Formato
Adobe PDF
|
8.76 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/10589/179512