Uncertainty characterization method for renewable energy and load based on fuzzy clustering

The integration of renewable energy sources into power systems poses significant challenges due to their inherent variability and uncertainty. Traditional deterministic methods are inadequate for addressing these stochastic characteristics, necessitating advanced probabilistic approaches. This paper presents a novel norm-based fuzzy clustering method for characterizing uncertainties in renewable energy and load outputs. We extend the traditional fuzzy C-means algorithm by incorporating L1 and L∞ norms alongside the L2 norm, and combine this with kernel density estimation to generate probability distribution functions of predicted power deviations. Case studies using real-world photovoltaic data demonstrate the method's effectiveness. Results show that our approach provides richer information through membership matrices and b etter handles datasets with fuzzy boundaries compared to K-means clustering. The proposed method offers a valuable tool for scenario generation and uncertainty characterization in power system planning and operation, contributing to improved integration of renewable energy sources.