Feature Extraction of Line Flow Fluctuation and Its Application to Dynamic Security Diagnosis

Abstract This paper proposes a method how to extract the features of line flow fluctuation and how to apply these features for system security monitoring. The feature extraction is done by a combined method of FFT processing and principal component analysis. This newly developed method is tested by using the field data acquired on a tie-line between the northeast area and the central area of Japan in 1987 and 1988. The result obtained from about 50 data sets shows that the proposed method is effective for extracting the features of the small fluctuation and the tieline flow fluctuation is characterized by three main features. Also, we discusses a concept of dynamic security diagnosis based on the proposed method. The statistical investigation shows that the most principal feature of the tie-line flow fluctuation is strongly related to the total generation level of our surveyed system. This fact suggests that the security diagnosis of knowledge based type will be realized by applying the relation between the extracted feature of line flow fluctuation and system condition.