Application of artificial intelligence based on state grid ESG platform in clean energy scheduling optimization.

The randomness and volatility of existing clean energy sources have increased the complexity of grid scheduling. To address this issue, this work proposes an artificial intelligence (AI) empowered method based on the Environmental, Social, and Governance (ESG) big data platform, focusing on multi-objective scheduling optimization for clean energy. This work employs a combination of Particle Swarm Optimization (PSO) and Deep Q-Network (DQN) to enhance grid scheduling efficiency and clean energy utilization. First, the work analyzes the complexity and uncertainty challenges faced in clean energy scheduling within the current power system, highlighting the limitations of traditional methods in handling multi-objective optimization and real-time response. Consequently, the work introduces the ESG big data platform and leverages its abundant data resources and computational power to improve the scheduling decision-making process. Next, PSO is used for initial scheduling optimization and a mathematical model for clean energy scheduling optimization is constructed. To further enhance the dynamic response capability of the scheduling process, this work designs a dual-layer architecture. Among this architecture, DQN is responsible for adjusting the PSO initially optimized scheduling plan based on real-time data during the actual scheduling process, adapting it to the instantaneous change demand and renewable energy output characteristics. Finally, to verify the model's effectiveness, this work conducts simulation analyses using real data from the state grid ESG big data platform. The results indicate that this method significantly improves clean energy utilization, increasing it from 62.4% (with traditional methods) to 87.7%, while reducing scheduling costs by 22%. With the increase in communication delay, the generation time of scheduling instruction increases significantly. This is because the algorithm needs to wait longer to receive the complete grid status information, which affects the generation speed of scheduling instructions. Moreover, this method demonstrates good adaptability and stability under different load demands and clean energy supply conditions. This work introduces a novel and efficient method for clean energy scheduling optimization by integrating PSO and DQN. It contributes to the overall improvement of clean energy utilization within the State Grid and provides a theoretical and empirical foundation for further research in related fields.