An Integrated MOPSO and Fuzzy Decision-Making Algorithmic Framework for Multi-Objective Configuration in High-Proportion Renewable Networks

The optimal sizing and configuration of energy storage in networks with high renewable penetration represent a highly complex, multi-constraint, and non-linear optimization problem. Existing planning methods often struggle with premature convergence and lack systematic multi-criteria evaluation for high-dimensional scenarios. This paper proposes a data-driven optimization and evaluation framework integrating a modified Multi-Objective Particle Swarm Optimization (MOPSO) algorithm and fuzzy mathematics based on deep learning. First, a time-series production simulation model is established to capture dynamic supply-demand interactions. The MOPSO algorithm is then introduced, utilizing a non-linear adaptive decreasing inertia weight and a Pareto dominance-based update strategy to effectively maintain population diversity and avoid local optima. Finally, to select the best compromise solution from the generated Pareto optimal solution set, a fuzzy membership evaluation algorithm is applied. Simulation on a 20 GW renewable base (16 GW wind + 4 GW PV) with 4 GW thermal regulation shows that a 6‑hour storage yields generation costs of 0.3002–0.4061 CNY/kWh. The proposed MOPSO outperforms NSGA‑II in convergence and solution quality, and fuzzy membership enables optimal trade‑off selection.