Hierarchical reinforcement learning based energy management strategy for hybrid electric vehicle

Qi, Chunyang; Zhu, Yiwen; Song, Chuanxue; Yan, Guangfu; Xiao, Feng; Wang, Da; Zhang, Xu; Cao, Jingwei

doi:10.1016/j.energy.2021.121703

Cited by 73 publications

(21 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, this approach requires advanced communication. The stability objective is accomplished using a non-droop method in the hierarchical controller, which implies that an additional controller is essential [49][50] [51].…”

Section: A Controlmentioning

confidence: 99%

Energy Management in Microgrid and Multi-microgrid

Xing¹,

Jia²

2023

Preprint

View full text Add to dashboard Cite

<p>Considered as basic structures of next-generation energy system, environment-friendly and flexible microgrid (MG) systems are potential solutions to deal with stochastic renewable energy sources (RESs). The essence of management is to maximize the benefits of available resources through coordination and optimization, which also applies to energy systems. Adaptable energy management approaches provide the possibility to construct effective and various energy interaction. The purpose of this paper is to present a systematic review of energy management in MG systems, and further to propose an energy management model. Therefore, this paper first comprehensively reviews recent research studies on MG, particularly in the interconnected mode (called as multi-microgrid, MMG). Based on the analysis of relative literatures, a STEER (stability, touch, efficiency, evenness, and resilience) strategy is proposed and formulated from the perspective of energy utilization for energy management. Then, critical technologies in energy management are reviewed and summarized from the aspects of control, communication, optimization, prediction, and evaluation. Furthermore, energy management system is illustrated from the perspectives of system function, management framework and operation logic, based on a four-layer hierarchical architecture. Eight main prospects on the future trend of energy management in MG and MMG are also presented.</p>

show abstract

Section: A Controlmentioning

confidence: 99%

Energy Management in Microgrid and Multi-microgrid

Xing¹,

Jia²

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…These vehicles provide propulsion from both an ICE using liquid fuel and an electric motor (EM). 6 Afterward, with the advancing battery technology, fully electric vehicles (FEVs) were developed. In these vehicles, the propulsion is provided by EMs fed from a DC energy source.…”

Section: Introductionmentioning

confidence: 99%

“…Hybrid electric vehicles (HEVs) were developed to reduce both carbon emissions and fossil fuel consumption. These vehicles provide propulsion from both an ICE using liquid fuel and an electric motor (EM) 6 . Afterward, with the advancing battery technology, fully electric vehicles (FEVs) were developed.…”

Section: Introductionmentioning

confidence: 99%

Double adaptive power allocation strategy in electric vehicles with battery/supercapacitor hybrid energy storage system

Çorapsız

Kahveci

2022

Intl J of Energy Research

View full text Add to dashboard Cite

Summary This paper proposes a new energy management strategy (EMS) for electric vehicles (EVs) with battery/supercapacitor hybrid energy storage systems (HESS). Firstly, the battery/supercapacitor HESS configuration and obtaining the load current from the driving cycle are comprehensively explained. Secondly, fixed and adaptive frequency‐based (AFB) EMS are discussed in detail. In the proposed method, the current demanded by the load is separated into low and high‐frequency components with the help of an adaptive low‐pass filter (A‐LPF). In addition, adaptive battery current (ABC) is generated according to the supercapacitor (SC) state of charge (SoC), and a double adaptive power allocation strategy is performed. The proposed method is compared with the AFB‐EMS for load currents, load powers, DC link voltages, filter cut‐off frequencies, battery, and SC SoCs on the UDDS driving cycle. When the obtained results were evaluated together, it was observed that the proposed method completed the driving cycle with a higher battery SoC and realized the adaptive cut‐off frequency in a narrower band.

show abstract

“…The reinforcement learningbased strategy can solve the optimal control problem in the case of no model or an inaccurate model, but the training of the neural network requires a complex configuration of hardware equipment and many iterations before convergence. 27 According to the above analysis, to design an EMS for practical application, the first point is to satisfy the optimality of energy optimization using vehicle dynamics information and optimization theory. Second, in the process of solving the optimal solution, the computational cost should be minimized to satisfy the requirements of the electronic control unit computational load.…”

Section: Introductionmentioning

confidence: 99%

“…To achieve the best fuel economy, maintain the battery state of charge (SOC), and slow fuel cell ageing, an adaptive online EMS based on the twin delayed deep deterministic policy gradient was designed in Reference [26]. The reinforcement learning‐based strategy can solve the optimal control problem in the case of no model or an inaccurate model, but the training of the neural network requires a complex configuration of hardware equipment and many iterations before convergence 27 …”

Section: Introductionmentioning

confidence: 99%

Adaptive equivalent consumption minimization strategy and its fast implementation of energy management for fuel cell electric vehicles

Cheng

Zhu

et al. 2022

Intl J of Energy Research

View full text Add to dashboard Cite

The equivalent consumption minimization strategy is widely used in real-time energy optimization for fuel cell electric vehicles due to its approximate optimality and good real-time performance. However, equivalent factors have always been difficult to select in the research of this algorithm, and the existing methods have insufficient adaptability and robustness. In this paper, an equivalent factor predictor based on a neural network is proposed, which can predict the equivalent factor in real time by considering various operating conditions and vehicle states. In addition, we improve the real-time performance of the equivalent consumption minimum strategy. A new method to quickly solve the continuous optimal solution of the algorithm by transforming the objective function into the quadratic function of the control variable is studied. Simulation and experimental results show that the designed equivalent consumption minimization strategy can ensure state of charge of the battery in a better range, and the energy saving effect is approximately 3% more than that of dynamic programming and 8% to 10% less than that of the traditional rule algorithm. Finally, the proposed method saves 2 to 3 orders of magnitude in computation time compared with the traditional equivalent consumption minimum strategy.

show abstract

Hierarchical reinforcement learning based energy management strategy for hybrid electric vehicle

Cited by 73 publications

References 25 publications

Energy Management in Microgrid and Multi-microgrid

Energy Management in Microgrid and Multi-microgrid

Double adaptive power allocation strategy in electric vehicles with battery/supercapacitor hybrid energy storage system

Adaptive equivalent consumption minimization strategy and its fast implementation of energy management for fuel cell electric vehicles

Contact Info

Product

Resources

About