Smart Energy Management for Electric and Hybrid Electric Vehicles

Personal transport is among the key contributors to global warming, whilst having significant importance for society. Hence, considerable efforts are made to make vehicles more sustainable. Different vehicle architectures, including batteries, internal combustion engines, super-capacitors, and fuel cell-based systems, coexist, including in the form of hybridization. These may require innovative fueling or charging solutions. Another key topic is energy management, which must be robust and well adapted for different driving situations. This article aims to provide an overview of the latest advancements in smart energy management for electric and hybrid electric vehicles, which were submitted to the Special Issue “Smart Energy Management for Electric and Hybrid Electric Vehicles” of the MDPI Journal Energies.

A comparative analysis of carbon dioxide emissions that are associated with different vehicle solutions provides an interesting starting point [1]. To contribute to the feasibility of the electrification of personal transport, the question of EV charging stations inside cities is addressed [2]. Moreover, among the existing vehicle architectures presented are series/parallel hybrid electric vehicles (HEVs) [3,4], plug-in hybrid electric vehicles (PHEVs) [5], and fuel cell hybrid electric vehicles (FCHEVs) [6].

Regarding energy management systems (EMSs), the presented solutions contribute to the need to establish robust EMSs that are capable of working in real time. The presented EMSs include a general approach [2], dual heuristic dynamic programming [3], the neural network-based approach [5], the reinforced learning-based approach [4], as well as a combination of optimization-based, rule-based, and learning-based strategies [6].

The aim of electrifying personal transport is to provide more sustainable solutions compared with the existing internal-engine-based approaches. Therefore, a comparative analysis of car engines using a sustainable approach, as presented by S. Grzesiak and A. Sulich in [1], is an important starting point. The authors compared “cars of the same producer, class and type with different engines including petrol, diesel, hybrid (petrol-electric), and electric engines in terms of environmental impact” [1]. This study considered the energy mix in Czechia, Germany, and Poland. “The result of [the article] indicates that vehicles with electric engines emit the least amount of carbon dioxide and are the most environmentally friendly solution in the given comparison criteria” [1].

One key to the electrification of personal transport is the availability of charging stations. An innovative solution regarding the implementation of a huge number of inner-city charging stations in cities using the trolley bus catenary grid is presented by M. Weisbach et al. in [2]. In this study, an intelligent multi-vehicle DC/DC solution for fast charging was designed using intelligent load management to meet the need for fast charging while respecting grid-level power limitations.

Hybrid solutions using two or more energy storage or conversion devices are among the methods to electrify personal transport. However, they require an EMS. As stated in [4], “the real-time control optimization of electrified vehicles is one of the most demanding tasks to be faced in the innovation progress of low-emission mobility”. In this context, M. Acquarone et al. produced their article titled “Influence of the Reward Function on the Selection of Reinforcement Learning Agents for Hybrid Electric Vehicles Real-Time Control” [4]. The authors used three reinforcement learning-based controllers (Q-learning, deep Q-Network, and double deep Q-Network) that outperform well-established real-time strategies on multiple driving missions with two reward functions: charge sustain and minimizing global fuel consumption. Their results show that the performances of the different reinforcement learning-based controllers are highly dependent on the power profile, especially if the systems are “trained on regulatory driving cycles and later tested on unknown real-world driving-missions” [4].

Dynamic programming (DP) is known to be capable of achieving a global minimum solution, while being computationally intensive and requiring prior knowledge of the entire driving cycle [5]. This is why D. Huo and P. Meckl [5] used a DP-based artificial neural network (ANN) approach to “get the benefit of using ANN to fit the DP solution so that it can be implemented in real-time for an arbitrary drive cycle” [5]. An intensive study of the hyperparameters’ effects showed that all “ANNs provide results that are comparable to the optimal DP solution” [5], while multiple-hidden-layer ANNs show better results than single-hidden-layer ANNs.

A similar idea was implemented by Y. Wang and X. Jiao in [3], where they used an adaptive dynamic programming (ADP)-based EMS for a series–parallel hybrid vehicle. The chosen dual heuristic dynamic programming (DHP) combines reinforcement learning with the DP optimization principle and a neural network-based approximation function [3]. A comparison with different control strategies showed a “robustness of fuel economy and the adaptability of the power-split optimization … to different driving conditions” [3].

In FCHEVs, the power split between the battery and fuel cell system (FCS) is also crucial. M. Matignon et al. proposed an integrated EMS concept, which used “the best of the three EMS categories (optimization-based (OBS), rules-based (RBS), and learning-based (LBS) strategies) to overcome the real-time operating condition limitations” [6]. In FCHEVs, the objective of control is to improve hydrogen consumption while managing the battery SOC under real-time driving conditions. Therefore, the authors used dynamic SOC horizon management, which was achieved by combining an RBS- with an LBS-based one to enhance the environmental data processing. The results show its ability to deal with real-time constraints whilst achieving an excellent performance compared with the optimal offline strategy.

The articles contributing to the MDPI Energies Special Issue “Smart Energy Management for Electric and Hybrid Electric Vehicles” demonstrate the progress in smart energy management for electric and hybrid vehicles, including contributions to plug-in HEVs, series/parallel HEVs, and FCHEVs. The authors made several contributions to meeting the demand for an EMS that is capable of working in real time while considering recent developments in control approaches, including a general approach [2], dual heuristic dynamic programming [3], a neural network-based approach [5], a reinforced learning-based approach [4], as well as a combination of optimization-based, rule-based, and learning-based strategies [6]. Moreover, questions regarding vehicle charging [2] and the sustainability of EVs and HEVs [1] are discussed.

The reader of this Special Issue can find a description of different vehicle types and different EMS approaches based on several driving cycles. The collected articles contribute to a better understanding of the needs of smart energy management for electric and hybrid electric vehicles.