This research introduces a novel intelligent protection technique for microgrids, leveraging Discrete Wavelet Transform (DWT) and an Ensemble Bagged Decision Tree (EBDT) classifier for precise fault detection and classification across various scenarios. Non-stationary voltage and current signals are analysed using DWT to extract wavelet detailed coefficients, which are then used to compute the energy of these coefficients as input features for the EBDT. The hyperparameters of the EBDT are optimized using a random search algorithm to enhance robustness in fault classification. The proposed protection scheme’s efficacy is validated on a modified IEC test microgrid model, incorporating both inverter-interfaced Distributed Generators (DGs) and synchronous DGs, under different fault conditions simulated in the MATLAB/SIMULINK environment. Results demonstrate that the method is both fast and accurate, achieving fault detection and classification accuracies of 100% in both grid-connected and islanded modes of operation. The performance of the proposed model is benchmarked against state-of-the-art methods, including Decision Tree and Random Forest classifiers. Additionally, the robustness of the proposed technique is confirmed under conditions of DG uncertainty and in the presence of noise. The effectiveness of the technique is assured across various fault conditions, and it is further validated in an OPAL-RT real-time environment. This study significantly enhances the reliability and resilience of microgrid protection systems, offering a robust solution for real-time fault management, which is crucial for the stability of modern power networks integrating renewable energy sources.
Read full abstract