Detection and classification of islanding by using variational mode decomposition and adaptive multi-kernel based extreme learning machine technique

In this work, a six bus IEC standard model is simulated under MATLAB/SIMULINK environment with different penetration levels of islanding i.e., small, medium, and large scenarios with radial as well as in looped configuration. The three-phase current signals are extracted from each end of buses and synthesized through variational mode decomposition (VMD) to compute the spectral energy. Thereafter, subsequent features are obtained and fed into a novel adaptive multi-kernel extreme learning machine (AMKELM) classifier to classify the islanding and non-islanding cases. Different switching conditions, low impedance, and high impedance faults, and interference of noise are purposefully taken into account by looking at the real-time conditions. Finally, the performance of the proposed technique is compared with other recently published research works as well as with traditional ones to prove its efficacy over other techniques for the detection and classification of islanding scenarios. (c) 2022 Elsevier Ltd. All rights reserved.