Protection Of Dc Ring Microgrid By Using Advanced Signal Processing And Machine Learning Techniques

Abstract

Nowadays, deploying fossil fuels, population growth, and the industrial revolution have newlineled to a global power crisis, prompting the immediate exploration of sustainable energy newlinesolutions. In this prospect, DC microgrids, especially renewable-based DC microgrids, newlineare emerging as a promising solution, offering increased efficiency, reliable operation, newlineless unit cost, etc. Among the various renewable energy sources, photovoltaic (PV) and newlinewind power offer substantial advantages and seamless integration with DC microgrids. newlineDespite their advantages, DC microgrids have inherent protection challenges that arise newlinefrom loose connections, load fluctuations, converter switching actions, and wearing newlinecable insulation, leading to potential overcurrent risks, instability voltage, various faults, newlineand islanding disturbances that need to be addressed prior to developing a DC microgrid. newlineTo address these challenges, conventional protection methods used in AC networks are newlinenot applicable for DC microgrids due to the absence of key parameters like frequency newlineand reactive power, as well as the lack of natural zero crossings within DC microgrids. newlineTherefore, proper protection strategies are crucial for uninterrupted power supply to the newlineconnected loads in DC microgrids during faults and various operating conditions. Thus, newlineto tackle these challenges and achieve robust performance in DC microgrids, we have newlineintroduced various approaches in this dissertation that encompass mathematical and newlinehybrid models. The mathematical-based approach realizes the fault current newlinecharacteristics and utilizes the difference current from the cable network for accurate newlinefault diagnosis. On the other hand, hybrid models encompass advanced signal processing newlineand machine learning techniques, and they are used for fault diagnosis, which includes newlinedetection, classification, and location estimation. newlineTo ensure accurate fault detection and enhance robustness in the context of the DC newlinemicrogrid, a detailed model is constructed in a MATLAB/Simulink environment. A newlinehybrid approach en