Haochen Hua

Demand-side energy storage and flexible loads are crucial for enhancing the stability and economy of microgrid operation. However, the integrated uncertainties and heterogeneous demand-side resources complicate coordinated scheduling, resulting in suboptimal utilization efficiency. This paper proposes a novel curriculum reinforcement learning architecture for collaborative scheduling of shared energy storage and flexible load. Flexible loads are constructed as virtual energy storage to increase the regulation capability of the system. A semi-coupled asynchronous optimization method based on reinforcement learning is proposed to solve the complex optimization scheduling problem. Furthermore, an adaptive curriculum learning method is designed to improve algorithm performance under uncertain environment by adaptively adjusting the difficulty of training tasks. The simulation results illustrate that the microgrid operator’s revenue obtained by the proposed energy scheduling method is significantly improved compared to the baseline methods, improving convergence speed by 34.92% and economy by 6.71% over the non-curriculum learning.

A large-scale microgrid cluster (MGC) using centralized energy management faces computational and data processing burdens, making it rather hard to obtain real-time optimal energy dispatch strategies. Network partitioning method can reduce such complexity by dividing the MGC into self-adequate partitions. However, existing dynamic network partitioning methods normally update the network partitioning results at fixed time intervals, which limits the ability to adapt to spatiotemporal variations in output of renewable energy sources (RESs), load demand, and grid topology. To address this, a dynamic network partitioning method based on an improved genetic algorithm (IGA) combined with depth-first search (DFS) is proposed. IGA is employed to optimize the partitioning index, and DFS is subsequently applied to obtain the network partitioning result. Furthermore, a dynamic update mechanism is designed to adaptively update the partitioning results according to spatiotemporal variations in system conditions. Simulation results on the modified IEEE 123-bus system show that the proposed method achieves better partitioning quality than the genetic algorithm (GA), spectral clustering, and Louvain algorithms while maintaining computational efficiency.

In the context of rapid growth in renewable energy installations and increasingly severe consumption issues, this paper designs a 100% green electricity supplied zero-carbon integrated energy station. It aims to analyze its configuration focusing on the following three core features: zero carbon emissions, 100% green electricity supply, and a centralized–distributed system structure. It discusses equipment selection and provides models for configuring upstream green electricity resources, power generation, energy storage, transformer, and energy conversion. The study examines the synergy between lithium-ion battery storage and modular molten salt thermal storage, along with the virtual energy storage characteristics formed by thermal load inertia, supported by mathematical models. Based on the data from a green electricity system in an Eastern Chinese city and typical load profiles, the paper validates a specific configuration for a 100% green electricity supplied zero-carbon integrated energy station, confirming model accuracy and calculating the required scale of upstream green electricity resources. It proves that establishing an electro-thermal storage synergy system is crucial for addressing the significant fluctuations in renewable energy output. It also argues that leveraging thermal load inertia to create virtual storage can reduce the investment in energy storage system construction.

Decarbonization of the whole energy chain has been recognized as a measure to tackle the global challenge of climate change, and significant progress has already been made on the generation side to integrate renewable energy. However, the demand side is the single largest underlying factor in shaping decarbonization roadmap. Hence, the carbon emission cost should also be shared by the users according to their power consumption. In this paper, a joint electricity-carbon trading framework is designed to reduce the carbon emission through trading and demand response. A delayed carbon emission liability settlement for asynchronous markets is proposed to ameliorate the users' optimal decision from single-point optimization to interval-based optimization. To develop the optimal strategy of trading within the proposed mechanism, an improved proximal policy optimization (PPO) algorithm based on Monte Carlo reward sampling is applied. Simulation studies reveal that, compared with the market without carbon trading and users without delayed settlement, the proposed mechanism has achieved a carbon emission reduction by 40.7% and 12.7% respectively. Simulations also show the algorithm's training efficiency can be significantly improved with the proposed Monte Carlo sampling method.