The climate-adaptive energy management system holds promising potential for harnessing the concealed energy-saving capabilities of connected plug-in hybrid electric vehicles. This research focuses on exploring the synergistic effects of artificial intelligence control and traffic preview to enhance the performance of the energy management system (EMS). A high-fidelity model of a multi-mode connected PHEV is calibrated using experimental data as a foundation. Subsequently, a model-free multistate deep reinforcement learning (DRL) algorithm is proposed to develop the integrated thermal and energy management (ITEM) system, incorporating features of engine smart warm-up and engine-assisted heating for cold climate conditions. The optimality and adaptability of the proposed system is evaluated through both offline tests and online hardware-in-the-loop tests, encompassing a homologation driving cycle and a real-world driving cycle in China with real-time traffic data. The results demonstrate that ITEM achieves a close to dynamic programming fuel economy performance with a margin of 93.7%, while reducing fuel consumption ranging from 2.2% to 9.6% as ambient temperature decreases from 15°C to -15°C in comparison to state-of-the-art DRL-based EMS solutions.

To develop an energy management strategy (EMS) that adapts to the state-of-health (SOH) of the fuel cell stack (FCS) to keep the output of the powertrain of fuel cell hybrid electric vehicles optimal all the time, a degradation adaptive EMS is proposed. The proposed EMS is based on a hierarchical framework, where the upper layer involves a degradation rate model to estimate the SOH of the FCS online and acquire performance characteristics at different SOH levels. In the lower layer, an EMS using the Rule-DDPG algorithm is developed. The results prove that the proposed method outperforms existing mainstream EMSs and aligns closely with the dynamic programming approach regarding economic and durability indicators. Compared to non-adaptive EMSs across various driving cycles, the proposed degradation adaptive EMS can consistently achieve the lowest energy consumption and optimal durability performance, tailored to the current SOH of the powertrain. In addition, a comparison between offline simulation and hard-in-the-loop testing validates the computational efficiency and real-time applicability of the proposed strategy in practical environments. Specifically, the calculation time for each sample during the WLTC cycle is shown to be less than 213.214ms, with an average calculation time of 168.637ms and a minimum calculation time of 150.618ms.

The advent of intelligent connected technology has greatly enriched the capabilities of vehicles in acquiring information. The integration of short-term information from limited sensing range and long-term information from cloud-based systems in vehicle motion planning and control has become a vital means to deeply explore the energy-saving potential of vehicles. In this study, a traffic-aware ecological cruising control (T-ECC) strategy based on a hierarchical framework for connected electric vehicles in uncertain traffic environments is proposed, leveraging the two distinct temporal-dimension information. In the upper layer that is dedicated for speed planning, a sustainable energy consumption strategy (SECS) is introduced for the first time. It finds the optimal economic speed by converting variations in kinetic energy into equivalent battery energy consumption based on long-term road information. In the lower layer, a synthetic rolling-horizon optimization control (SROC) is developed to handle real-time traffic uncertainties. This control approach jointly optimizes energy efficiency, battery life, driving safety, and comfort for vehicles under dynamically changing traffic conditions. Notably, a stochastic preceding vehicle model is presented to effectively capture the uncertainties in traffic during the driving process. Finally, the proposed T-ECC is validated through simulations in both virtual and real-world driving conditions. Results demonstrate that the proposed strategy significantly improves the energy efficiency of the vehicle.

The intricate interactions with other road users and the diversity of traffic environments create a challenging decision-making task for autonomous driving systems. While offline learning solutions are renowned for their high execution efficiency and the ability to approximate the optimal policy across the entire state space, they are often unsafe and fragile when encountering untrained states. Conversely, online planning methods possess the capacity to thoroughly assess how current decisions influence future outcomes online, and therefore have better generalization. Nevertheless, these approaches face difficulties in terms of execution efficiency and are susceptible to becoming stuck in locally optimal solutions. In this context, this paper proposes an Integrated Planning and Learning (IPL) algorithm based on the reinforcement learning framework for speed and lane change decision-making on highways. Specifically, at each decision time step, this method utilizes the offline learned model to guide an online Monte Carlo Tree Search (MCTS) algorithm for heuristic search, aiming to formulate a forward-looking policy. The experimental results show that the IPL algorithm performs better generalization when faced with unknown scenarios, and its asymptotic performance is better than other benchmark algorithms. In addition, in contrast to the MCTS-based online planning method, the IPL algorithm enhances execution efficiency and comes closer to achieving global optimality.

Automotive energy management systems (EMSs) are evolving to achieve intelligence across their entire lifecycle, from initial product development to real-world customer usage. This paper proposes a bi-level transfer approach with model-agnostic meta-learning (MAML) to realize cross-platform transferable and online-adaptive EMS. In the development phase, MAML calibrates the heuristic control maps of an instantaneous optimization-based EMS, with a per-unit state-action space design ensuring seamless knowledge transfer between vehicle platforms. During the usage phase, real driving data are adopted to refine onboard control parameters. The framework is validated through real vehicle experiments. Firstly, the entire MAML-assisted V-cycle development is implemented to validate the optimality and knowledge transfer of the EMS, resulting in zero-shot transfer for EMS calibration on new vehicle products. Additionally, real vehicle experimental tests show that a correction of 8.0%~9.5% fuel economy is improved against the convention reinforcement learning-based EMS during usage via online-adaptation, effectively bridging the gap between pre-trained policies and real-world optimal energy management.