The shared energy storage device acts as an energy hub between multiple microgrids to better play the complementary characteristics of the microgrid power cycle. In this paper, the cooperative operation process of shared energy storage participating in multiple island microgrid systems is researched, and the two-stage research on multi-microgrid operation mode and shared energy storage optimization service cost is focused on. In the first stage, the output of each subject is determined with the goal of profit optimization and optimal energy storage capacity, and the modified grey wolf algorithm is used to solve the problem. In the second stage, the income distribution problem is transformed into a negotiation bargaining process. The island microgrid and the shared energy storage are the two sides of the game. Combined with the non-cooperative game theory, the alternating direction multiplier method is used to reduce the shared energy storage service cost. The simulation results show that shared energy storage can optimize the allocation of multi-party resources by flexibly adjusting the control mode, improving the efficiency of resource utilization while improving the consumption of renewable energy, meeting the power demand of all parties, and realizing the sharing of energy storage resources. Simulation results show that compared with the traditional PSO algorithm, the iterative times of the GWO algorithm proposed in this paper are reduced by 35.62%, and the calculation time is shortened by 34.34%. Compared with the common GWO algorithm, the number of iterations is reduced by 18.97%, and the calculation time is shortened by 22.31%.

The promotion of local renewable energy consumption and stable power gird (the latter is referred to as PG) operation have emerged as the primary objectives of power system reform. The integration of multiple microgrids with distinct characteristics through the utilization of shared energy storage (the following is referred to as SES) facilitates coordinated operation. This approach enables the balancing of energy across temporal and spatial domains, contributing to the overall reliability and security of the energy network. The proposed model outlines a methodology for the coordinated operation of multiple microgrids and SES, with a focus on asymmetric price negotiation. Initially, cost and revenue models for microgrids and SES power plants are established. Secondly, an asymmetric pricing method based on the magnitude of each entity’s energy contribution is proposed. A profit optimization model is also established. The model can be decomposed into two distinct subproblems: the maximization of overall profit and the negotiation of transaction prices. The model can be solved by employing the alternating direction method of multipliers (ADMM). Finally, a series of case studies were conducted for the purpose of validating the operation optimization model that was previously constructed. These studies demonstrate that the model enhances collective operational efficiency by 44.69%, with each entity’s efficiency increasing by at least 12%. At the same time, cooperative benefits are distributed fairly according to each entity’s energy contribution.

Green low-carbon data centers operate as coupled cyber-energy systems whose dispatch must coordinate renewable generation, grid exchange, battery storage, cooling load, flexible computing workload, carbon-intensity signals, and reliability constraints. This study develops and evaluates a safety-constrained deep reinforcement learning framework for source–load–storage coordinated operation of a grid-connected green data center. The operating problem is formulated as a constrained Markov decision process with state variables describing the IT load, deferrable workload backlog, renewable availability, electricity price, marginal carbon intensity, battery state of charge, server-room temperature, reserve margin, and calendar context. The action space covers grid import and export, renewable utilization, storage charge and discharge, workload shifting, and cooling control. The learning architecture combines a constrained actor–critic policy, adaptive Lagrangian safety critics, and a control barrier function (CBF)-based action shield that projects unsafe actions onto an explicitly defined operating set before plant execution. The shield is specified as a low-dimensional quadratic projection over state-dependent SOC, thermal, reserve, SLA, and grid-interface constraints, while cumulative risks are priced through Lagrangian safety budgets during policy training. The evaluation uses a controlled and auditable benchmark simulation with normalized public-data-compatible profiles, declared scenarios, random seeds, neural-network settings, and mechanism-matched baselines; it is not a telemetry-based verification or hardware certification of a deployed data center. Within this declared benchmark, the proposed safe DRL controller produces a simulated 13.1% emission reduction relative to the Rule-based controller, 95.8% renewable utilization, a normalized annual cost of 0.91, and fewer boundary contacts than the tested unconstrained, Lagrangian-only, and shield-only PPO variants. These percentages are simulator outputs relative to the stated benchmark and must not be interpreted as measured field savings. The results show how separating reward learning, cumulative safety pricing, and one-step engineering projection changes low-carbon dispatch within the specified model.