Power distribution abnormal data identification and processing system based on correlation classification

Abstract when the conventional system classifies power distribution data, data category is not fine enough, resulting in low accuracy of abnormal identification, high false detection rate and large correction error. A identification processing based on correlation classification proposed. In terms hardware, multi-layer framework adopted, overall hardware structure composed communication modules, so as to optimize signal acquisition circuit suppress noise; software, set convergence conditions, iteratively increase number clustering centers until conditions are met, realize draw classified curve, obtain normal feasible region, judge that there abnormalities beyond use vertical comparison method for correction. The results show designed improves reduces error, better than system.