Empirical orthogonal decomposition applied to transmission line protection

In protective relaying, reducing the fault clearance time is always desired for ensuring transient stability and also for reducing damage on the equipment. With this purpose, protection schemes based on the analysis of off-nominal frequencies are being investigated. Most proposals on this field use sine-cosine decomposition techniques for spectral analysis of transient signals. However those techniques fail to provide simplified information since results spread in a wide frequency range. In this work, a representative sample of fault signals are used to find a small set of orthogonal functions which gives a better interpretation concerning fault generated transients. Results show that fault signals decomposed in terms of these orthogonal basis functions exhibit well-defined patterns, which can be used for recognizing the main features of fault events such as inception angle, fault type and fault location. 2 ms data windows are used, thus the methodology could be applied in high-speed protection schemes.