On-line disturbance classification using nearest neighbor rule

This paper presents an automated on-line disturbance classification technique for different power quality problems. This technique is based on wavelet multi-resolution analysis and nearest neighbors pattern recognition method. The wavelet-multi-resolution transform is introduced as a powerful tool for feature extraction. It has the ability to extract discriminative, translation invariant features with small dimensionality in order to classify different disturbances. The nearest neighbor pattern recognition technique is then implemented to classify different disturbances and evaluate the efficiency of the extracted features.