Riassunto analitico
In the context of motorcycle, we can assist to an increasing interest toward semi-active suspension control systems able to improve both the comfort and the passenger’s safety in
both racing and original equipment manufacturer applications. Such systems implement suitable strategies based on the measure of several quantities, among which the relative
position and velocity of the wheels respect to the vehicle body with the aim of regulating in real-time the damping forces. The actual effectiveness of such strategy strongly depends
on the reliability and accuracy of the data measured by the sensors involved in the control loop. Due to their simplicity and good performance in terms of linearity, the most used
sensors for suspension displacement measurements are based on linear potentiometers but such kind of sensors is expensive for mass market and unreliable in the long run.
In this thesis an innovative approach is exploited, based on the application of machine learning algorithms, to give the basis for the design and validation of a soft sensor for
the on-line estimation of the suspensions travel of the Ducati Panigale V4 from signals available onboard on the CANbus.
Main features of the core model as well as the corresponding simulated results are discussed with reference to experimental data, collected during motorcycle track tests,
which are used to train a regression algorithm for future prediction. Different architectures are tested and compared to obtain an accurate and efficient model, which presents an
accuracy of about 5% with a confidence of 95%. In particular, a black box modeling approach is deployed and compared to a gray box one which requires more information
and parameters to capture the underlying physics of the dynamic system.
|
Abstract
In the context of motorcycle, we can assist to an increasing interest toward semi-active suspension control systems able to improve both the comfort and the passenger’s safety in
both racing and original equipment manufacturer applications. Such systems implement suitable strategies based on the measure of several quantities, among which the relative
position and velocity of the wheels respect to the vehicle body with the aim of regulating in real-time the damping forces. The actual effectiveness of such strategy strongly depends
on the reliability and accuracy of the data measured by the sensors involved in the control loop. Due to their simplicity and good performance in terms of linearity, the most used
sensors for suspension displacement measurements are based on linear potentiometers but such kind of sensors is expensive for mass market and unreliable in the long run.
In this thesis an innovative approach is exploited, based on the application of machine learning algorithms, to give the basis for the design and validation of a soft sensor for
the on-line estimation of the suspensions travel of the Ducati Panigale V4 from signals available onboard on the CANbus.
Main features of the core model as well as the corresponding simulated results are discussed with reference to experimental data, collected during motorcycle track tests,
which are used to train a regression algorithm for future prediction. Different architectures are tested and compared to obtain an accurate and efficient model, which presents an
accuracy of about 5% with a confidence of 95%. In particular, a black box modeling approach is deployed and compared to a gray box one which requires more information
and parameters to capture the underlying physics of the dynamic system.
|
