On-line disturbance recognition utilizing vector quantization based fast match

In this paper, a given power quality event is matched against a set of known power quality events in a database. Matlab was utilized in constructing this database by simulating different power quality phenomena. Feature vectors are extracted from windowed Fourier transform or windowed Walsh transform. These features are further processed by vector quantization for minimizing the calculation burden. A fast match (FM) algorithm is proposed to find the degree of similarity between the labels of tested phenomena and the pre-stored labels for different power quality phenomena. In this paper we study the effect of changing the codebook size, and type of feature extraction used on the classification accuracy of the proposed classifier. Results show that the implementation of such a classifier is simple, yet it results in high classification rate, which makes it a very competitive alternative for already existing classifiers, which suffer from large computation burden and complicated structure.