Communication‐based predictive energy management strategy for a hybrid powertrain

Abstract

Predictive information can significantly improve energy efficiency in a hybrid powertrain, especially for a long drive. Ideally the prediction of a sufficiently long horizon can bring the maximum benefit, however during this horizon the traffic can change, making it impossible to continue the chosen optimal strategy. Due to the limited vision of vehicles' sensors, it is difficult to acquire information far away. Against this background, this article proposes a predictive optimal control strategy in the connected environment. It combines real-world vehicle-to-everything (V2X) information and cloud communication to allow for swift reaction and loss mitigation. More specifically, V2X information is used to detect impending changes on the route and the cloud is used to map these changes onto new constraints on the operation of the HEV. Instead of using "static" V2X information directly, a more realistic prediction method considering the dynamic of traffic is developed. Besides, an update strategy is adopted to timely cope with uncertainties. The performance of the approach is shown by a case study based on real driving cycles in Austria. The results show that the proposed structure is able to improve the performance (mainly fuel efficiency) up to 6.9%.

A comparative study of driver torque demand prediction methods

“…Another emerging scenario that has been addressed in the literature is the vehicle-to-everything (V2X) context, where the vehicle is connected to the surrounding environment and it is able to exchange information with any entity that can affect its performance and energy consumption. For example, Deng et al [29,30] have proposed a model-free approach and a multi-layer architecture to predict long-distance traffic velocity in real-world V2X considering the future dynamics of V2X information. The prediction has been based on the most similar past trajectories to the given route and this information has been exploited for forecasting up to 20 min; then, traffic velocity predicted has been adopted for an MPC control approach.…”

“…For example, Deng et al. [29, 30] have proposed a model‐free approach and a multi‐layer architecture to predict long‐distance traffic velocity in real‐world V2X considering the future dynamics of V2X information. The prediction has been based on the most similar past trajectories to the given route and this information has been exploited for forecasting up to 20 min; then, traffic velocity predicted has been adopted for an MPC control approach.…”

Optimal control of networked and transportation systems

“…A scenario-based optimized controller tuning is proposed to maximize the throughput while accounting for the current traffic situation and, therefore, optimize the infrastructure utilization [4]. A double-layer predictive control for hybrid powertrain, using currently available vehicle-to-everything information and cloud computing, is developed and assessed on a real case study [5].In summary, this special issue provides a fresh view on the most recent developments in optimal control of networked and transportation systems, by considering theory, algorithms, and applications.…”

“…A second group of papers [4,5] considers transport systems applications of networked optimal control theory. A scenario‐based optimized controller tuning is proposed to maximize the throughput while accounting for the current traffic situation and, therefore, optimize the infrastructure utilization [4].…”

“…A scenario‐based optimized controller tuning is proposed to maximize the throughput while accounting for the current traffic situation and, therefore, optimize the infrastructure utilization [4]. A double‐layer predictive control for hybrid powertrain, using currently available vehicle‐to‐everything information and cloud computing, is developed and assessed on a real case study [5].…”