Load characteristics clustering of dynamic modeling data

Load current data can be regarded as stochastic disturbance of voltage. Three decompositions and re-construction of wavelet package method are be used to analyze load modeling data, and characterizes vector of load data are accurately constructed which is used to classify load data. The characteristics vector was standardized, then load data was classified by fuzzy subtract clustering in the paper. The proposed method has been proved validity by the examples of the attaining characteristics and clustering of dynamic lab and transformer substation data, which is high precision and convergence. It has the significant to load modeling processing the amount of load data.