A novel wind power equivalent method based on clustering of multivariable panel data

With the rapid development of large clusters of wind power, a multi-machine dynamic equivalent model reflecting accurately wind speed diversity and correlations between wind generators should be established to study the influence of multiple spatio-temporal characteristics of wind power on power system. In this paper, dependent variable index characteristics and their correlations with independent variable indexes of wind power are considered from time and cross-section in two-dimensional space by introducing panel data analysis theory. Building the similar and related distance measure function of wind power panel data as well as combining with cuckoo search algorithm, a novel wind power dynamic equivalent method is proposed based on the clustering of multivariable panel data. The detailed model and aggregate model are built by using DIgSILENT/PowerFactory software. The results demonstrate that the proposed cluster classification indexes are effective, which are able to reflecting accurately actual characteristics of the wind farm.