Classification of Underlying Causes of Power Quality Disturbances Using Data Fusion

Abstract

This paper develops a data fusion technology based modeling framework for classifying the underlying cause of power quality (PQ) disturbances. First, the moving-window technique is used to cluster disturbance period with the consideration of the temporal propagation of disturbance energy. Secondly, the PQ disturbance measurements, equipment switching action data and alarm events are integrated by utilizing entity matching method. Then, the distributed mining of association rules is designed to obtain strong association rules within integrated data for describing the relationship between PQ features and event causes. The analysis results have good generalization performance. Finally, the real grid data were taken as an example to verify the effectiveness and practicability of the proposed method. The test results show that the proposed method can analyze the relationship between the typical PQ disturbance features and event causes effectively. This relationship is meaningful for power quality improvement.