The exhaust ultimate power valve system is an electro-mechanical system composed by an electric motor, a couple of cables and mechanical valve. The valve, electronically controlled through the motor, is placed in the exhaust, generally after the catalyst and its main objective is to allow the motorbike phonometric homologation. This is basically done lengthening the exhaust gases path and therefore reducing the motorbike noise when it is needed. Nevertheless, it has several other aims to pursue, such as acting as a low-rpm boosting tool.
For this thesis work all the tests and measurements have been performed through a HIL test bench. It is composed by the exhaust ultimate power hatch valve of the Ducati Panigale V4, two Bowden cables, the electric motor Kokusan - KL1623, supplied by a 12 V power supply through a harness, and the dSPACE MicroAutoBox II. The latter is a development control unit for performing fast function prototyping.
The initial part of the work has been focused on the identification of the system transfer function through an experimental method. It has been chosen to consider the Duty Cycle [%] as input, while the Speed [mV/s] as output. The first outcome has been the need of two different transfer functions, one for positive and one for negative movements. This is due to the the presence of a spring on the valve pulley that turns out in a nonlinearity of the system. Indeed, to close the valve, it is necessary to contrast the elastic force of the spring, while, when an opening movement is imposed, the elastic force of the spring helps the motion.
After this preliminary observation, the measurements of the system response to the scheduled step and sinusoidal inputs have been performed, as well as the necessary data post-processing to import them in System Identification Toolbox. Once all the data have been acquired, the model order of the system has been estimated, based on a rough knowledge of the system physic. It has been chosen to model the system under study with a second order system, comparable to a mass-spring-damper one. Therefore, the selected transfer function has two poles and no zero.
After that, it has been thought to characterise gain and poles through different types of acquisitions. Indeed, the estimation of the transfer function gain has been made from the step acquisitions, due to the clear linearity between applied inputs and measured outputs. Whereas, the identification of the transfer function poles has been performed from the sinusoidal measurements, because the wide range of frequencies investigated has shown a couple of complex conjugate poles in a limited position in terms of frequency.
Alongside the transfer function identification, the electronic control strategy has been created on Simulink. The aim of the strategy is to control the position and the speed of the valve through the application of a suitable duty cycle. The position target set by the ECU has to be achieved, while, respecting the constraints on Speed [mV/s], Current [mA] and Position [mV], and detecting both electrical and mechanical faults.
The basic structure of the strategy is composed by four different operating modes, a supervisor stateflow machine and an errors diagnosis stateflow machine. The four operating modes have been called ”Precalibration”, ”Calibration”, ”Keep Alive” and ”Position Control”: among them, only one at a time can operate. The decision of which one has to be executed is taken by the supervisor stateflow machine. The latter machine operates in parallel with the errors diagnosis one, in which around twenty different errors are constantly checked, depending on the operating mode. In the case of an error detection, the strategy provides for disabling the control of the valve, supplying a null Duty Cycle [%] to the electric motor.
Once the system transfer function has been identified, it has been exploited to properly calibrate the PI controller.