An Improved Method for Classifying Power Quality Disturbances

An adaptive neuro-fuzzy inference classifier based on the discrete wavelet transform (DWT) to recognize the type of power quality (PQ) disturbances is presented. The DWT, using the multi-resolution signal decomposition (MSD), can transfer power disturbance characteristics into the time-frequency domain. The energy of the signal decomposed to frequency sub-bands can be used to extract feature parameters for classifying various disturbances. The proposed classifier was designed using four feature parameters that consist of energy concentration level and its mean value, mean energy of the signal, and an auxiliary parameter determined by the rms value and pulse detection. The proposed classifier shows good recognizing efficiency for ten types of PQ events, including one double event disturbance.