High impedance fault detection in medium voltage distribution systems using wavelet transform

Abstract

The increasing complexity of distribution systems demands innovative approaches to fault detection, particularly for high impedance faults which are characterized by low fault currents that often elude conventional detection methods. This thesis addresses this challenge by focusing on the development of a novel and reliable algorithm for high impedance fault detection in medium voltage distribution systems. The methodology employed is pivoted on wavelet transform analysis, a powerful tool for signal processing and analysis, to enhance fault detection capabilities. Wavelet transform provides the analysis of different frequency components of a signal with good time and frequency localization. The main objective of this research is to address the limitations of existing high impedance fault detection methods by proposing a new algorithm that offers improved reliability and accuracy. The proposed algorithm leverages wavelet transform to carry out a multi resolution analysis using 'sym5' as a mother wavelet to extract level 5 detail coefficients from the electrical current signals of the distribution system. These coefficients capture subtle variations of high frequency components which are indicative of high impedance faults. By calculating the energies of the level 5 coefficients for each phase and the neutral, and subsequently integrating the results to determine the average slope at each phase and neutral, a robust metric for fault detection is established. The algorithm's effectiveness is validated through extensive testing, with results consistently aligning with expectations. The testing phase involves simulated fault scenarios using MATLAB/Simulink software to assess the algorithm's performance under various conditions. The outcomes demonstrate the algorithm's ability to accurately detect high impedance faults and the faulty phase(s), showcasing its potential as a valuable tool for enhancing the reliability of medium voltage distribution systems. In conclusion, the research presents a successful algorithm for high impedance fault detection, offering a promising solution to a critical challenge in power distribution systems. The proposed method exhibits a level of reliability that surpasses traditional approaches, laying the foundation for improved fault detection strategies. However, recognizing the complexity and variability of real-world distribution systems along with the difficulty in modeling high impedance faults perfectly, further testing on real world situations is recommended to validate the algorithm's performance under diverse operating conditions and in the presence of additional system complexities. To ensure the algorithm's practical applicability, it is essential to conduct field tests in real-life distribution systems. These tests will provide insights into the algorithm's adaptability to varying environmental conditions and its ability to accommodate the nuances of different distribution network configurations. Additionally, the algorithm's performance in the presence of multiple simultaneous faults and its response to transient conditions warrant investigation to assess its robustness. Recommendations for future research include expanding the algorithm's capabilities to address more complex fault scenarios and exploring opportunities for integration with emerging technologies such as machine learning for enhanced fault detection and classification. Continued collaboration with industry stakeholders and power system operators is crucial to fostering the adoption of this innovative algorithm and ensuring its seamless integration into practical applications. In summary, this thesis contributes a significant advancement in high impedance fault detection, offering a reliable algorithm grounded in wavelet transform analysis. The potential impact of this research extends to the improvement of power system reliability and resilience, safeguarding against faults that pose challenges to traditional detection methods. Through ongoing research, testing, and collaboration, the proposed algorithm holds the promise of becoming an indispensable tool in the ongoing evolution of medium voltage distribution systems.