The objective of this paper is to recognize the power quality events (PQEs) by wavelet transform (WT) and categorization by modified extreme learning machine (MELM). The nature of power quality events is non-stationary and discrete wavelet transform (DWT) by multiresolution analysis (MRA) is used to analyze those signals. In this approach, applying the WT on all the spectral components, the typical features of PQ event signals have been obtained and on noisy conditions in order to analyze the performance of the proposed method, three types of PQ event data sets are constructed by assembling noise of 25dB, 35dB and 45dB. For generalized single hidden layer feedforward networks (SLFNs), modified ELM is an efficient learning algorithm which is implemented to recognizing the various PQEs classes. Under ideal and noisy conditions based on very high performance, the proposed WT-MELM method has robust recognition structure that can be used in real power systems.

disturbance Shannon entropy index, wavelet transform (WT), modified extreme learning machine (MELM), nonstationary power signal.