Design and development of a control algorithm to damp drive-train oscillations in multi engine vehicles (Diplomarbeit)


Matthias Hammer


Abstract - This thesis deals with the feedback control of drive-train oscillation in Hybrid vehicles. The drive-train system consists of a pair of real wheels and a pair front wheels. In the case of a hybrid vehicle, each pair of wheel is powered with one or more engines - either internal combustion engine (ICE) or electric motor (EM). Depending on a particular input (e.g., sudden change in speed) from the driver, the torques are generated by the engines and often a sudden change in such torques causes an oscillation in vehicle dynamics. Such oscillation is undesirable to maintain a comfort level for the occupants of the vehicle. This project aims towards designing a suitable feedback control strategy to reduce/avoid the above mentioned drive-train oscillation. Development and implementation of a feedback control algorithm for a given dynamic system, in general, has three distinct phases. First, it is necessary to identify the system dynamics. The outcome of this phase is a set of differential equations which models the behavior of the actual physical system as closely as possible. Next task to come up with a suitable control algorithm to achieve a desired set of system behavior. Finally, the designed algorithm needs to be implemented on a an hardware architecture. In automotive domain, the implementation architecture is often networked and one needs to decide on (i) task partitioning (ii) task mapping (iii) bus communication schedule and their impact on the control algorithm (iv) depending on the architecture there might be a need for redesigning the control algorithm. In the above context, this thesis has three main contributions: (i) system identification for the drive-train oscillation in hybrid vehicle (ii) control algorithm design to reduce the oscillatory movement (iii) implementation on a real networked architecture (i.e., that are used in BMW hybrid vehicles) consisting of multiple electronic control units (ECUs) with legacy tasks and test the designed control algorithm.