Robust Dynamic Modeling and Optimal Scheduling of Wind-Solar-Storage Systems via Multi-Modal Data Fusion Under Uncertainty

The dynamic modeling and scheduling optimization of wind-solar storage systems have attracted much attention with the growing share of wind and solar energy in the power system. But the variability of wind and solar power generation as well as the complex configuration of energy storage systems create challenges to this process. To improve the robustness of the wind-solar-storage system against uncertainty in a dynamic manner, this paper proposes a Method of multi-mode data fusion based scheduling optimization. We create a multimodal data fusion framework by combining meteorological data (wind speed, light intensity, temperature, etc.), historical power generation data, and on-site monitoring data. The deep learning models enable the framework to gather multiple datasets and later learn to improve the system's predictive ability. Meanwhile, from the perspective of scheduling optimization, a scheduling decision optimization method based on the Deep Q Network is used to dynamically adjust wind and solar power generation and energy storage strategies according to the real-time system state, to optimize scheduling decision, so as to balance economic benefits and system stability. The experimental results indicate that the proposed dynamic modeling and scheduling optimization method based on multi-modal data fusion is effective for dealing with the uncertainty of wind-solar-storage system, and achieves better scheduling efficiency and stability of the system.