This paper provides a comprehensive review of hybrid energy systems (HESs), focusing on their challenges, optimization techniques, and control strategies to enhance performance, reliability, and sustainability across various applications, such as microgrids (MGs), commercial buildings, healthcare facilities, and cruise ships. The integration of renewable energy sources (RESs), including solar photovoltaics (PVs), with enabling technologies such as fuel cells (FCs), batteries (BTs), and energy storage systems (ESSs) plays a critical role in improving energy management, reducing emissions, and increasing economic viability. This review highlights advancements in multi-objective optimization techniques, real-time energy management, and sophisticated control strategies that have significantly contributed to reducing fuel consumption, operational costs, and environmental impact. However, key challenges remain, including the scalability of optimization techniques, sensitivity to system parameter variations, and limited incorporation of user behavior, grid dynamics, and life cycle carbon emissions. The review underlines the need for robust, adaptable control strategies capable of accommodating rapidly changing energy environments, as well as the importance of life cycle assessments to ensure the long-term sustainability of RES technologies. Future research directions emphasize the integration of variable RESs, advanced scheduling, and the application of emerging technologies such as artificial intelligence and blockchain to improve system resilience and efficiency. This paper introduces a novel classification framework, distinct from existing taxonomies, addressing gaps in prior reviews by incorporating emerging technologies and focusing on the dynamic nature of energy management in hybrid systems. It also advocates for bridging the gap between theoretical advancements and real-world implementation to promote the development of more sustainable and reliable HESs.

The increasing of grid-connected variable renewable energy (VRE) on the demand side causes balance problems in the power system. Thus, dealing with the uncertainty, variability, and consequently, flexibility requirement is becoming an urgent challenge to the power system operators. Virtual power plant (VPP), which bundles different types of small-scale distributed energy resources (DERs) into a single unit for optimization will effectively mitigate those uncertainties. An optimal VPP energy management method is proposed in this article for optimal energy and operating reserve (OR) scheduling. The studied VPP is a cluster of dispersed generating units (including dispatchable and stochastic power sources), flexible loads, as well as storage units. VPP operator has to make decisions based on the uncertainty coming from the stochastic VRE, load demand, as well as market electricity price. Thus, a dynamic risk reserve quantification method is proposed to cover both VRE power and load forecast uncertainties, while information gap decision theory is applied in the unit commitment procedures to study the impact of price uncertainty on the decision-making of VPP operators. Finally, the proposed method is implemented and verified with a case study, and optimal decisions are discussed.

This paper proposes a novel energy management approach (imitation-Q-learning) based on imitation learning (IL) and reinforcement learning (RL). The proposed approach reinforces a decision-making agent based on a modified Q-learning algorithm to mimic an expert demonstration to solve a microgrid (MG) energy management problem. Those demonstrations are derived from solving a set of linear programming (LP) problems. Consequently, the imitation-Q-learning algorithm learns by interacting with the MG simulator and imitating the LP demonstrations to make decisions in real time that minimize the MG energy costs without prior knowledge of uncertainties related to photovoltaic (PV) production, load consumption, and electricity prices. A real-scale MG at the Lille Catholic University in France was used as a case study to conduct experiments. The proposed approach was compared to the expert performances, which are the LP algorithm and the conventional Q-learning algorithm in different test scenarios. It was approximately 80 times faster than conventional Q-learning and achieved the same performance as LP. In order to test the robustness of the proposed approach, a PV inverter crush and load shedding were also simulated. Preliminary results show the effectiveness of the proposed method.