Enhanced Fault Diagnosis in AC Microgrids Using Adaptive Wavelet Packet Decomposition and Decision Tree Classifier

With the increasing adoption of carbon-neutral energy and the demand for reliable and eco-conscious power systems, Microgrids (MGs) have emerged as a key solution to modern energy challenges. By integrating diverse energy sources and operating in both grid-synchronized and islanded (self-sufficiently) modes, MGs enhance energy efficiency, resilience, and sustainability. However, their complex structure and dynamic operational conditions present significant challenges in fault identification and protection. Traditional methods often fail to address these challenges due to limitations in adaptability and real-time responsiveness. This paper proposes a novel approach combining Adaptive Wavelet Packet Decomposition (AWPD) with a Decision Tree (DT) classifier to enhance fault identification and categorization in AC microgrids (MGs). AWPD effectively extracts features from non-stationary and transient signals, while the DT classifier offers high accuracy and interpretability in fault categorization. Comparative analysis with other ML (machine learning) models demonstrates the superior performance of the proposed method, achieving over 99% accuracy in classifying distinct fault types. This study underscores the potential of integrating advanced signal processing with ML techniques for intelligent microgrid protection and lays a foundation for future exploration in real-time implementation and broader system applications.