A Study on the Fuel-Saving Potential of a Multi-Criteria Speed Trajectory Planning for HEV

The automation of longitudinal vehicle dynamics is one of the major challenges of today’s automotive development and is of great importance for driver assistance systems and autonomous driving. In hybrid electric vehicles, this automation offers great potential for saving fuel due to the additional degree of freedom in the powertrain. To minimize fuel consumption, it is of central importance how to link the driving strategy and the operating strategy of the hybrid electric powertrain. A possible solution for this is offered by the multicriteria driving strategy planning presented in [1], which computes fuel-efficient speed trajectories for following other road users, where the planning is restarted after a certain distance (driving horizon). In addition to [1] the aim of the paper at hand is to quantify the fuel-saving potential of this speed trajectory planning with a study using real data of different route types (urban, extra-urban, combined, highway) and emission test cycles such as the WLTC. For this purpose, design of experiments was used to identify the parameters that have a decisive influence on the fuel-saving potential. The multi-criteria driving strategy planning is based on an algorithm that uses a parameter variation of the car-following model Intelligent Driver Model (IDM) to compute many possible speed trajectories to follow the driver ahead for a certain driving horizon. The most suitable trajectory is then selected by means of a cost function and the operating strategy is rescheduled with this trajectory. The operating strategy can be integrated into the algorithm in different ways: The operating strategy can be rescheduled only for the current driving horizon, or from the current position to the end of the trip – which is called the local, respectively global approach of rescheduling. Furthermore, the stochastic method of a Sequential Importance Resampling particle filtering algorithm can be used to determine the speed trajectory instead of simply considering all possible trajectories in a driving horizon, which greatly reduces computing time. To quantify the fuel-saving potential of the multi-criteria speed trajectory planning, it is investigated which of the two operating strategy integration approaches shows better fuel consumption results, which route type provides the highest fuelsaving potential and whether the use of the particle filtering algorithm has a negative effect on the fuel-saving potential. The fuel-saving potential is defined here as the reduction of fuel consumption by using the multi-criteria driving strategy planning when following another road user compared to the use of a car following model with static standard driving parameters as defined in [2]. To set up the study, the system understanding was first deepened by an algorithm analysis. Moreover, improvements in the parametrization of the cost function, the IDM and the particle filtering algorithm were made and a new standardized approach was created as a reference for the study. Results of the design of experiments identified the length of the driving horizon, the weighting of energy and time efficiency in the trajectory planning and the use of the particle filtering algorithm as the most influencing parameters for the fuel-saving potential. In summary, the results show that the use of multi-criteria driving strategy planning to plan speed trajectories for hybrid electric vehicles offers a general fuel-saving potential and that the maximum potential is 9,75 %. Furthermore, the global approach is 1,57 % more fuel efficient than the local approach of rescheduling the operating strategy. Overall, urban driving profiles showed the highest fuel-saving potential of all possible route types. The use of the particle filter algorithm reduces the fuelsaving potential by 0,89 % compared to the simulation with all trajectories, because it calculates only an approximate optimal speed trajectory with respect to the cost function. However, this stochastic method is currently the only way to achieve real-time capability for road use. Finally, the most profitable driving horizon was found to be 100 m in length for planning the speed trajectory. The paper concludes with a recommendation for the parameterization of the algorithm and further steps for implementation in a real vehicle.