Outlier Identification and Justification Using Multi-Objective PSO based Clustering Algorithm in Power System

A clustering method based on multi-objective PSO is developed in this paper for bad data identification. The algorithm automatically determines the optimum number of clusters and it starts by partitioning the data set into a relatively large number of clusters to reduce the effects of initial conditions. After the bad data are detected, eigencurves of correlating load extracted by Kohonen network are used to modify the bad data. The application of the proposed clustering algorithm to the problem of unsupervised classification of electric load data is investigated. Through the simulation test, the results show the effectiveness of the algorithm.