Automatic segmentation method for voltage sag detection and characterization

Abstract

Although characterization of voltage sag is an essential part of voltage sag studies, the way that taking magnitude and duration as acknowledged basic characteristics cannot describe sag characteristics versus time. Hence automatic segmentation, which divides monitoring data sequence into segments, and characterization algorithm are proposed in this paper. The difficulty that how to divide segment automatically is overcome through two-stage segmentation algorithm based on singular value decomposition method. Then multi-dimension characteristics such as magnitude, duration, phase-angle jump, sag type and so on can be calculated. Hundreds of sag events data including measured in field and synthetic are utilized to validate the effectiveness and reliability of proposed method. Moreover, the detection and characterization algorithm are ported to installed monitors and backstage data center with C programming, timesaving and practical get validated.