Predictive longitudinal control offers a flexible solution for reducing fuel consumption in motor vehicles, which does not require any changes to components. The reduction is therefore determined solely by the more intelligent operation, based on a model of the system and knowledge about the route ahead. Various methods of optimal control can be used to optimize this system. In this thesis different direct methods and dynamic programming are presented. In contrast to dynamic programming, direct methods cannot deal directly with non-continuous optimization variables. These discrete optimization variables must first be converted into a continuous problem, for example by convexification. The methods of optimal control can then be used in the context of a model-based predictive control in the vehicle. To determine the fuel consumption of vehicles with combustion engines and the energy consumption of battery electric vehicles, an additional consumption model of the drive is required. The Willans Line Model offers an approximation for many drive components. The fuel savings of the predictive longitudinal control depend on various factors. On the one hand, this depends on the implementation of predictive longitudinal control, but also on the vehicle itself. To analyze these influences, the fuel consumption for different parameters is first determined using dynamic programming. Especially the maximum and minimum decelerations in the deceleration phases as well as the use of coasting have a great influence. A linear model of the vehicle parameters as a function of the vehicle mass is used to investigate the influence of different vehicle classes. In this way, the fuel savings can be determined for different vehicle classes. Especially for heavier vehicles, the predictive longitudinal control has a positive effect. With battery electric vehicles, the energy saving is less pronounced due to the high efficiency of the drive over wide operating ranges. The implementation of the predictive longitudinal control can be carried out with the optimization methods presented. Due to the low number of system states and the discrete decision variables of coasting, it is advisable to implement this using dynamic programming. Especially if the costs are calculated in advance, a longitudinal control with very low computing time requirement can be developed. Different temporal constraints can be introduced by extending the dynamic programming. Thus time restrictions, as they arise when a minimum average speed is required, can be taken into account by multi-criteria dynamic programming. Time is considered as an additional cost term in the vectorial cost function. Time constraints, e.g. traffic lights or other road users, can be mapped by considering time as a further state of the system. In both cases, this converts the previously scalar optimization problem into a multi-criteria optimization problem, which increases the computing requirements. Under the assumption of constant acceleration processes, however, the temporal constraints can also be mapped in the velocity band. This means that scalar optimization can be applied again, but the solution space is severely restricted, resulting in higher consumption compared to the optimal solution.     «

Predictive longitudinal control offers a flexible solution for reducing fuel consumption in motor vehicles, which does not require any changes to components. The reduction is therefore determined solely by the more intelligent operation, based on a model of the system and knowledge about the route ahead. Various methods of optimal control can be used to optimize this system. In this thesis different direct methods and dynamic programming are presented. In contrast to dynamic programming, direct...     »