Performance assessment of load profiles clustering methods based on silhouette analysis

The paper presents a performance assessment of load profiles clustering methods using silhouette value criterion, a method of interpretation and validation of consistency within clusters of data. Three of the most common clustering methods are considered: Density-Based Spatial Clustering of Applications with Noise, Hierarchical cluster analysis, k-means clustering. Based on a large dataset of real medium voltage (MV) substation load profiles, the approaches have been first assessed on multiple smaller subsets of randomly selected data. Different representations of the data are considered in terms of temporal resolution and data scaling varying clustering parameters in a sort of sensitivity analysis. Based on average silhouette values, the clustering methods are ranked. The approaches are then applied to the entire dataset and, based on the clusters identified, some standard load profiles are extracted, shown, and briefly discuss.