Analysis of islanded microgrid using principal component analysis

Abstract

In the Indian scenario, rural electrification is considered an important newlinegovernment policy that has the off-grid distribution system as an entity. newlineConsidering the autonomous nature of off-grid distribution systems, smart newlinegrid implementation is inevitable. Smart Grid Environment evolves with both newlinemeasurement and communication technologies to collect and process the newlineheterogeneous big data. The higher the size of the data, higher is the newlinecomputational complexity of the data processing. Finding means to reduce the newlinecomputational complexity and data size becomes a significant aspect in newlinedeveloping the smart grid environment. Given the sophistication of power newlineelectronics technologies, penetration of distributed generation in the grid newlineenvironment and off-grid environment has shown unprecedented progress. newlineThe stability analysis of any complex power system like the off-grid parallel newlineinverters involves numerous state variables. Data explosion in the smart grid newlineenvironment due to more variables reckons more hardware space and cost due newlineto the higher memory usage and computational space. newlineThis research is an attempt to reduce the computational complexity newlinein the steady state analysis of parallel inverters, using machine learning newlinealgorithms in a smart off-grid environment using feature reduction methods. newlineParallel inverters are self-synchronizing inverters that do not need the newlinereference from the main grid to synchronize between them. Voltage and phase newlineangle are measured for the inverters using the droop characteristics. newlineDroop curves of instantaneous angular frequency versus real power newlineand voltage versus instantaneous reactive power act as the lookup table to get newlineinstantaneous voltage and phase angle. newline