This paper brings forward a new method to classify the cases of power quality. The fractal exponent wavelet analysis is adopted to obtain the whole exponents and the local exponents of the disturbance waveform. Two steps are included in the method: one divides the cases into three groups based on the whole exponents and the other redivides two of the three groups by the fast clustering algorithm based on reference and density (CURD). The parameters after simple pretreatment of the local exponents act as the points in the clustering. CURD captures the shape and extent of a cluster by references and analyzes the points based on the reference. CURD also keeps the ability of density. 240 cases, 12 types of disturbances emulated by PSCAD/EMTDC are included in the application. The method has the potential to distinguish the test data.
展开▼
