TY - JOUR TI - An energy-efficient scheduling approach for wind-solar-hydrogen systems based on distributed reinforcement learning JO - AI Agent PY - 2026 VL - 2 IS - 2 SP - EP - 21 SN - ISSN 3070-3719 (Online) AB -

This paper presents a comprehensive energy dispatch strategy based on distributed reinforcement learning to optimize the operation of integrated wind-solar-hydrogen systems. The proposed approach effectively reduces coal fuel costs and carbon emissions while ensuring precise load demand tracking. By implementing a distributed computing framework, the computational challenges associated with training the Deep Deterministic Policy Gradient algorithm on large-scale datasets are effectively addressed. This parallel architecture significantly enhances training efficiency and improves scalability for complex energy management tasks. Additionally, an efficient load pattern identification method, enhanced by Principal Component Analysis and K-means clustering, is developed to capture the salient characteristics of electricity load data. Furthermore, a high-fidelity representative scenario extraction approach, utilizing Dynamic Time Warping and Density-Based Spatial Clustering of Applications with Noise, is proposed to characterize the inherent uncertainties in wind and solar power generation. The integration of hydrogen-based energy storage is proposed as a flexible and sustainable solution to enhance system reliability and mitigate carbon emissions. Empirical simulation results demonstrate that the proposed methodology significantly reduces fuel costs and minimizes carbon emissions while exhibiting improved robustness and computational efficiency. By incorporating hydrogen storage systems and carbon trading mechanisms, the proposed approach optimally facilitates the integration of wind and solar power, thereby providing a comprehensive framework for the efficient operation of hybrid energy systems.

KW - Carbon trading KW - multi-objective optimization KW - distributed reinforcement learning KW - wind-solar-hydrogen systems DO - 10.20517/aiagent.2026.01 UR - https://dx.doi.org/10.20517/aiagent.2026.01