Control of a brake by wire actuator for racing motorbikes

This thesis deals with the control of an innovative Brake By Wire (BBW) actuator particularly designed for racing motorbikes. The final goal of a BBW actuator is to control the torque at the vehicle wheel, however in this application, due to cost, encumbrance and reliability constraints an indirect torque control is performed regulating the pressure in the brake caliper. In this application the control requirements are strict: absence of overshoot and bandwidth greater than $10$ Hertz are required. This, together with the intrinsic system time-variant non-linearity makes the control problem not trivial. The starting point of the control design is the system model: writing each actuator component equations, a physical based complete model is derived. This model represents very well the system experimental response and it is useful to adopt it as a simulator; however it is too complex to be employed for control design purposes. For this reason, besides the complete model, a control oriented one is derived. This model provides a sufficient system description for control purpose; its analysis gives important guidelines for the design of the control algorithm. Once the system to be controlled is modeled, the control problem can be faced; in this thesis it is solved following two approaches. A first one is focused on exploring innovative and advanced control techniques; these control techniques are studied and then adapted to the particular system, finally they are validated on a test bench. Pursuing this approach simplifies the testing phase: in fact, validating the control algorithms on a test bench allows us to discard the safety requirements, which are really strict and severe in this application. This is not possible when testing the control algorithms on a real system, i.e. on a racing motorbike. The second approach is focused on solving the control problem adopting conventional control techniques; in this case the control algorithm are studied and designed based on the system model, then they are implemented and tested on a real motorbike. For this reason, in parallel with the control design, the fault detection algorithm design performs an important role. In pursuing the first approach, firstly we deal with the friction, which has a dominant role in the BBW actuator. In this context the classical dithering compensation is compared with a more sophisticated model based compensation approach. In particular, the model based approach approximates the non-linear friction model with an innovative linear-in-the-parameters (LP) approximation, this is updated through an adaptation mechanism. In parallel to the adaptive friction compensation technique, a sliding mode based controller guarantees the pressure tracking. Then, anti-windup compensation technique is employed in order to build a compensator able to linearize the system intrinsic non-linearity. At this point, considering that in real systems control action has an upper physical saturation due to hardware limitation, an anti-windup compensation is derived. The employment of these two compensators permit to design a pressure control on a linear system, without considering the intrinsic system non-linearities. Moreover, this control algorithm does not discard the control action saturation in the design phase. Finally, the Virtual Reference Feedback Tuning (VRFT) paradigm is experimentally validated on the test bench; in doing this we focused on a particular task that it could be necessary when designing the control algorithm: the design of a position controller. In this particular case of study we compared a canonical model based PID tuning with the VRFT approach, showing that the VRFT algorithm provides satisfactory performance. This approach makes the control design faster and avoids critical phases such as the identification one and the controller designed one. Following the conventional control techniques and exploiting the peculiarities that the control oriented model highlights, we propose two different control strategies: an adaptive position-pressure switching control and an adaptive cascade position-pressure. These control strategies are simple, robust and easy to tune. Their design phase is performed on the test bench, then they are implemented and tested on a real motorbike, showing satisfactory performances. Moreover, due to the safety critical application, a fault detection algorithm suited for the particular case of study is presented.

Questa tesi tratta il controllo di un attuatore frenante (BBW) per motoveicoli da competizione. L'obbiettivo di controllo e' quello di imporre una coppia frenante alla ruota; siccome per motivi di costo, ingombro ed affidabilita' il sistema non e' equipaggiato con un sensore di coppia, in questo la variabile da controllare e' la pressione. In questa particolare applicazione, i requisiti di controllo sono molto esigenti: assenza di sovraelongazioni e una banda in anello chiuso di almeno 10 Hz. Questo, insieme alle non linearita' intrinsiche del sistema, rendono il problema di controllo non banale. Il punto di partenza per lo sviluppo delle strategie di controllo proposte e' la modellistica dell'attuatore: partendo dalle equazioni fisiche dei componenti che compongono il sistema, un modello completo viene presentato. Questo modello, nonostante sia molto dettagliato e' fedele alle realta', e' troppo complesso per essere impiegato per fini modellistici. Percio', partendo da esso e facendo delle assunzioni, un modello piu' semplice che descrive solo le dinamiche di maggior interesse viene ricavato. Once the system to be controlled is modeled, the control problem can be faced; in this thesis it is solved following two approaches. A first one is focused on exploring innovative and advanced control techniques; these control techniques are studied and then adapted to the particular system, finally they are validated on a test bench. Pursuing this approach simplifies the testing phase: in fact, validating the control algorithms on a test bench allows us to discard the safety requirements, which are really strict and severe in this application. This is not possible when testing the control algorithms on a real system, i.e. on a racing motorbike. The second approach is focused on solving the control problem adopting conventional control techniques; in this case the control algorithm are studied and designed based on the system model, then they are implemented and tested on a real motorbike. For this reason, in parallel with the control design, the fault detection algorithm design performs an important role. Il problema di controllo e' risolto seguendo due metodologie: una e' piu' metodologica e' consiste nell'usare l'attuatore come banco di prova per strategie di controllo avanzate ed innovative. L'altra invece sfrutta tecniche di controllo classico. In questo caso le tecniche di controllo vengono testate anche su una motocicletta da competizione, quindi affianco ad esse, e' stata sviluppata una raffinata fault detection in grado di garantire la sicurezza del motociclista.