An Application of Multidimensional Scaling in Fault Detection of Smart Grids

Abstract

Monitoring smart grids and detecting faults in such huge networks has recently become an active area of research. The huge amount of data transferred from the measurement units to the control center makes it difficult to detect faults in a reasonable amount of time. Some existing methods have been investigated and tested on many IEEE models such as wavelet transform, principal component analysis (PCA) to extract the abnormal behavior of signals under monitoring [1]. However, such techniques pose difficulties in detecting different faults properly. In this paper, the multidimensional scaling (MDS) is investigated as an alternative technique for reducing the dimensionality of the data to lower dimensions, while maintaining the necessary information needed for fault detection. MDS is then used to investigate the behavior of some IEEE models under different types of faults in order to detect and locate the faulty bus bars.