Abstract: This paper describes predictive modeling applied to optimization of solar power generation systems. Such modeling, based on machine learning principles, is performed for both solar irradiation and load demand, applied to both redistribution of load demand to specified time slots, and to time-specified prediction of power generation and load demand. Weather prediction is the most important part of solar power generation forecasting, particularly with reference to solar resource inflation, deflation, and backfill. A method of optimization of solar power generation systems combining known methods is proposed. AI-enhanced predictive modeling, neural fuzzy modeling with fuzzy-weighted regionalization, net-load creation and solar power generation forecasting from multi-analysis of previous generation time history, event overlay on prediction of net-load shape, prediction combining, envelope based backfilling, and bottoming by thermal and hydro resources, are elements used.
Generalization of predictive modeling principles and methods, in particular for net-load modeling, can be performed for any other renewable power source. Electricity load demand forecasting is one of the most challenging tasks in power distribution system management, in both near and long terms. For forecasting, the main challenge consists in the presence of some characteristic load structures such as daily and even weekly periodicities, promotion for special events, season and long period past generalization by means of supporting production of particular events similarly defined, high relationship of non-shiftable elements on near and mid-term forecasts, and season relationships to long-term ones. There are two different approximation intents, and accuracy somewhat split between them.
Keywords: AI-driven optimization, solar power generation, predictive weather modeling, load forecasting, machine learning, energy management, renewable energy, power output prediction, real-time data, smart grid, energy efficiency, photovoltaic systems, deep learning, demand prediction, energy forecasting, intelligent control systems, data analytics, operational efficiency, weather-based optimization, sustainable energy systems.
|
DOI:
10.17148/IJARCCE.2022.111254
