This thesis project deals with the study of a particular type of reactive suspension which enables the elastic force modulation through the opening and closure of a valve placed in between two connected chambers of a multichamber air-spring. Two models, of different complexity, that exhaustively represent its behavior are developed. Once the different physical and structural parameters are estimated through the definition of an optimization problem, a validation is performed by confronting the simulated signals with the ones experimentally obtained on a test bench. The simpler model is used to define an offline optimizer that, known the trend of a specific road profile, is able to determine the optimal switch sequence of the valve, improving in this way the driver’s comfort. Subsequently, based on the algorithm’s choices, a neural network is trained with the capability of replicating, at least partially, the performances reached by the optimizer. Moreover, the network is capable of working in real-time, therefore without knowing the road profile in advance.
Questa tesi verte sullo studio di un tipo di sospensione reattiva che permette la modulazione della forza elastica attraverso l’apertura e la chiusura di una valvola posta tra due camere comunicanti di una molla ad aria multicamera. Vengono sviluppati due modelli, di differente complessità, che rappresentano in modo esaustivo il suo comportamento. Quindi, stimati i vari parametri fisici e strutturali attraverso la definizione di un problema di ottimizzazione, vengono validati con un confronto dei segnali simulati con quelli ottenuti sperimentalmente su un banco prova. Il modello più semplice viene utilizzato per la sintesi di un ottimizzatore offline che, noto l’andamento di uno specifico profilo stradale, trova la sequenza ottima di switch della valvola migliorando il comfort del guidatore. Basandosi poi sulle scelte prese dall’algoritmo viene istruita una rete neurale che risulti capace di replicare, almeno in parte, le performace raggiunte dall’ottimizzatore, lavorando questa volta in real-time.
Analysis and control of a multichamber suspension
DI PALMA, FEDERICO
2018/2019
Abstract
This thesis project deals with the study of a particular type of reactive suspension which enables the elastic force modulation through the opening and closure of a valve placed in between two connected chambers of a multichamber air-spring. Two models, of different complexity, that exhaustively represent its behavior are developed. Once the different physical and structural parameters are estimated through the definition of an optimization problem, a validation is performed by confronting the simulated signals with the ones experimentally obtained on a test bench. The simpler model is used to define an offline optimizer that, known the trend of a specific road profile, is able to determine the optimal switch sequence of the valve, improving in this way the driver’s comfort. Subsequently, based on the algorithm’s choices, a neural network is trained with the capability of replicating, at least partially, the performances reached by the optimizer. Moreover, the network is capable of working in real-time, therefore without knowing the road profile in advance.File | Dimensione | Formato | |
---|---|---|---|
2020_04_DiPalma.pdf
non accessibile
Descrizione: Testo della tesi
Dimensione
5.27 MB
Formato
Adobe PDF
|
5.27 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/10589/165224