Run feature extraction of current amplitude domain in new energy power generation system

Driven by the “double carbon” target, the power grid presents a typical double-high characteristic of high proportion of renewable energy and high proportion of power electronic equipment. The increasingly complex dynamic signals in the power grid make the current meter out of tolerance, which seriously affects the accuracy of electric energy measurement. In order to extract and characterize the characteristics of fluctuating current, Based on EMD empirical mode decomposition, this paper extracts the fundamental amplitude and quasi-steady-state term of the grid-connected current. Secondly, the quasi-steady-state term is used as the interception curve to intercept the dynamic power signal, and the Canopy algorithm is used to coarsely cluster the run-length sequence to obtain the optimal clustering number K and the initial clustering center. The run-length sequences are finely clustered by Kmeans and Kshape clustering methods respectively, and the obtained two sets of clustering results ar e characterized and compared. The intra-class compactness and inter-class separation of the two clustering results are measured and evaluated, and the Kshape clustering result is better.