A Novel Clustering Method for PV Power Curve Patterns based on Multidimensional Feature, Entropy Weight, and K-means

• Published in: Engineering letters Vol. 33; no. 4; p. 876
• Main Authors: Li, Xingzhen, Ma, Yiwei, Zhong, Hao, Huang, Miao
• Format: Journal Article
• Language: English
• Published: Hong Kong International Association of Engineers 01.04.2025
• Subjects: Algorithms; Clustering; Entropy; Entropy (statistics); Euclidean geometry; Factor analysis; Photovoltaic cells; Weighting methods
• ISSN: 1816-093X; 1816-0948

The clustering of photovoltaic (PV) power is a challenging issue, as it is subject to various natural meteorological factors. To address this issue, this paper proposes a novel clustering method for PV power curve patterns based on multidimensional feature, entropy weight method, and K-means algorithm. First, a multidimensional feature model is proposed to better reveal the PV power characteristics, which integrates three major power fluctuation features and four major meteorological features based on factor analysis (FA). Second, the entropy weight method (EWM) is adopted to calculate the weight for each feature, which is used to modify the Euclidean distance in the K-means algorithm for high-quality performance of PV power curve clustering. The experimental results show that this method is more effective than traditional methods in terms of clustering indicators and clustering quality, as it achieves the best results in SC, DBI, and CHI clustering validity indices of 0.4463, 1.0981, and 393.9127, respectively.