Modification of the Sugar-James Algorithm for Clustering Measurements of Mode Parameters

The paper proposes the modified Sugar-James algorithm for identifying the number of clusters in measurements of mode parameters. This algorithm is based on the maximization of jumps of transformed intra cluster dispersion due to setting the dependence of the transformation degree from the number of clusters. The number of clusters selection is performed by an analysis of transformation degree dependence from the number of clusters. The dependence provides a maximum jump of transformed distortions. The results of the testing the algorithm by test measurements, following the normal distribution law are given, as well as comparison with the results of the Sugar-James algorithm other versions, where a constant value or a value, inversely proportional to the number of clusters, is given as the transformation degree. The modified algorithm is tested by real measurements of power mode parameters at load node. The results of the testing and verification by real measurements show the correctness of the number of clusters identification, and, also, the ability to identify the electrical load states.