Exploring Neural Networks for Enhanced Grid Stability and Intelligent Energy Distribution in SolarPowered Homes

Abstract

Artificial neural networks are broadly considered particularly well suited for implementation in smart grid environments as they can be utilized to treat various advanced applications in a more general and adaptable manner than traditional algorithmic approaches. Recognizing this benefit, a body of research has been dedicated to exploration of use of artificial neural networks within smart grid settings. These applications range from forecasting aspects in renewables planning, advance energy pricing and information modeling purposes to security issues and reinforcement learning applications. Furthermore, neuromorphic algorithms for the smart grid have also been considered, as they can capture convenient network topology of grid systems, and can potentially treat spatio-temporal analysis of the grid system itself, thus providing valuable diagnostics in certain smart grid subproblems. Interest concerning residential and architectural applications moves research largely toward these objectives. The microgrid concept implanted in dwellings raises intelligent appliances and control of distributed generation, thus benefiting from artificial neural networks flexible and adaptable control schemes. Moreover, investment in renewable energy generation malware increases in households, thus expanding power electronics applications. Artificial neural networks can successfully treat this complex nonlinear system. This shows a clear need for a master’s dissertation covering this topical field, summarizing applications of artificial neural networks in a smart grid context and addressing their potential impact both in electric power engineering and residential applications. A literature review is provided of the state of the art features smart grid elements as well as background knowledge on artificial neural networks to prepare the reader for proposing new intelligent systems in this field. Subsequently, objectives and methodology of the experimental work are described. Artificial neural network platforms are utilized in creation of intelligent systems for 1) enhanced grid stability through frequency estimation, 2) intelligent sensing and microphone arrays for monitoring of grid systems, and 3) intelligent energy distribution in solar powered homes. A dataset of this experiment is analyzed and visualized for exploration of its characteristics, thus establishing a methodology for appropriate network design and providing a basis for further estimations and intelligent systems.