Development environments for route based functions in road vehicles

Abstract

The electrification of heavy vehicles has increased demands on accurate range estimation to support efficient transport planning and reduce range anxiety. Consequently, route-based predictive energy estimation has become an important tool for improving the reliability of range predictions. The energy consumption of heavy vehicles is strongly influenced by factors such as vehicle mass, route topography, traffic conditions, and driving behaviour. This thesis investigates a predictive range estimation development environment for battery-electric heavy vehicles using a Kalman filter-based energy estimator integrated with TruckMaker simulations and MATLAB-based post-processing. The estimator combines route, weather, and mission information to predict battery energy consumption, state of charge (SOC), and remaining driving range together with its associated variance along a predefined route between the Port of Gothenburg and Viared. The accuracy of the estimator was evaluated by comparing predicted vehicle behaviour with TruckMaker simulation results. In addition, the influence of re-estimation frequency, vehicle mass uncertainty, and traffic disturbances was investigated. Several distance-based re-estimation intervals were analysed to evaluate how periodic updates affect prediction accuracy. Furthermore, scenarios involving incorrect vehicle mass information, mid-route mass correction, and traffic congestion were introduced to assess the robustness of the prediction framework under changing operating conditions. The results show that the estimator is capable of predicting SOC and vehicle range with good accuracy under nominal conditions. Periodic re-estimation significantly improves prediction accuracy and enables the estimator to adapt to changing operating conditions throughout the route. Vehicle mass was identified as one of the most influential parameters affecting prediction accuracy, while the investigated traffic disturbance had a comparatively smaller impact on the overall energy consumption. The driver-related estimator parameters K and σv,ref were also investigated. Both parameters influenced the predicted mean energy consumption and the associated uncertainty, while updating them to values identified from previously driven data resulted in only minor changes in the SOC prediction. These results indicate that the parameters may be useful for tuning the predicted mean energy demand and its associated uncertainty. Overall, the results indicate that continuous re-estimation can improve the robustness of route-based range prediction for battery-electric heavy vehicles.