Vol.:(0123456789) 1 3 Arabian Journal for Science and Engineering https://doi.org/10.1007/s13369-019-04289-5 RESEARCH ARTICLE - -ELECTRICAL ENGINEERING Power Quality Events Recognition Using S‑Transform and Wild Goat Optimization‑Based Extreme Learning Machine Indu Sekhar Samanta 1  · Pravat Kumar Rout 2  · Satyasis Mishra 1 Received: 26 July 2019 / Accepted: 3 December 2019 © King Fahd University of Petroleum & Minerals 2019 Abstract This paper presents a novel approach for automatic power quality (PQ) event detection and classifcation based on Stock- well transform (S-transform) and wild goat optimization (WGO)-tuned extreme learning machine (ELM). The distinctive features associated with PQ event signals have been extracted by S-transform to obtain the feature vectors characterizing the signal nature. Considering these feature vectors as input, a classifer based on ELM optimally tuned with modifed WGO technique is proposed. The WGO technique originated from the social hierarchy and strategic planning to reach at peak by the wild goats in nature is adapted to formulate an efective ELM model by parameter tuning for better classifcation. To justify the enhanced performance of the proposed approach, it is tested on a wide range of extracted synthetic PQ event data by MATLAB simulation. To ensure the real-time implementation, the PQ event data with the addition of 20, 30, and 50 dB to the synthetic signals are considered. The analysis of results presented reveals a very high performance for both PQ event recognition and classifcation, ensuring the efciency of the proposed approach. Keywords Time–frequency analysis · Extreme learning machine · Power quality · Feature extraction · Classifcation · Parameter tuning · Wild goat optimization List of Symbols w(d, t) Width of the wavelet d Scale parameter f Frequency S(t, f) Stockwell transform h(kT) Disturbance signal T Sampling time interval N Number of samples n Number of input neurons l Number of hidden neurons m Number of output neurons h Hidden layer output G(x) Hidden layer activation function a Weight matrix between input neuron and hidden neuron β Weight matrix between input neuron and hidden neuron b Weight of hidden neuron bias H Output matrix of the hidden layer H + Moore–Penrose inverse of the matrix H wg ij Population matrix N wg Number of wild goats N var Number of variables WT Weight of each wg N l Number of leaders N f Number of followers N g Number of groups GR Group of wild goats vv Movement vectors P lbest Best leader wight value w Inertia weight R Personal learning coefcient c, d Auxiliary parameters wg Wild goat k Index of the group WT G Group weight value m Mutation percentage m’ Ratio of the current-generation iteration number to the maximum iteration number MP i Total number of misclassifed patterns CL Class f s Sampling frequency * Indu Sekhar Samanta indusekhars5@gmail.com 1 Department of Electronics and Communication Engineering, Centurion University, Odisha, India 2 Department of Electrical and Electronics Engineering, Siksha ‘O’ Anusandhan University, Bhubaneswar, India