Random Matrix Theory-Based Data-Driven Framework for Demand Response

With the increasing integration of renewable energy, microgrids face challenges in maintaining grid stability due to variability and uncertainty in energy supply. Enhancing microgrid flexibility through demand response is essential, but accurately assessing demand response potential is complex due to high-dimensional and noisy data. This paper proposes a framework combining Random Matrix Theory (RMT) for high-dimensional data analysis and a Deep Auto-Encoder Network (DAE) for feature extraction and visualization. RMT identifies hidden correlations between microgrid users and environmental factors, optimizing demand response strategies. The DAE reduces data dimensionality, providing clear visual insights into user behavior and demand response potential. A case study based on data from a southern city in China demonstrates the effectiveness of the proposed method. Results show seasonal variations in demand response potential, highlighting the need for tailored strategies. This approach enhances situational awareness and supports renewable energy integration in microgrids.