Non stationary signal analysis and classification using FTT transform and Naive Bayes classifier

The paper proposes a fast Time-Time transform for pattern recognition and analysis of non stationary signal which are responsible for power quality (PQ) distortion. The Time-Time transform is generalized and a faster algorithm is devised for faster implementation. To distinguish among various power disturbances statistical features were extracted from Time-Time distribution. These statistical features are then passed through an efficient classifier namely Naive Bayes classifier for identification of signal patterns. Improved performance can be clearly demonstrated through a classification accuracies comparison between earlier proposed techniques.