Energy storage optimization method based on coordinated control of power balance and microgrid

With the increasing demand for power balance and collaborative control in microgrids, relying on deep reinforcement learning algorithms to construct a computerized and efficient control framework, and achieving optimized scheduling of energy storage systems, has become a key direction for solving the problem of economic stability in microgrids with a high proportion of renewable energy access. Therefore, a research proposes an energy storage optimization model that integrates deep learning and reinforcement learning algorithms. This model improves the accuracy and adaptability of energy storage scheduling through global optimization using Particle Swarm Optimization (PSO) algorithm, continuous action processing of deep deterministic strategy, and strategy update stability of proximal strategy optimization. In the microgrid datasets of San Diego and Mesadel Sol in California, compared with three traditional models, this model has a minimum convergence iteration of 31 times and the highest energy storage state of charge control accuracy of 98.22%. This model improves the performance of microgrid power balance and collaborative control, providing an efficient solution for optimizing and scheduling energy storage systems.