Libros y Capítulos de libros

URI permanente para esta colección

https://repositorio.uft.cl/handle/20.500.12254/2438

Examinar

Examinando Libros y Capítulos de libros por Materia "Machine learning"

Mostrando 1 - 2 de 2
  • Miniatura
    Ítem
    Forecasting Electric Power Generation in a Photovoltaic Power Systems for Smart Energy Management
    (Institute of Electrical and Electronics Engineers (IEEE), 2022-08-25) Rao, Challa Krishna; Sahoo, Sarat Kumar; Yanine, Fernando
    Solar electricity is generated using photovoltaic (PV) systems all over the world. Solar power sources are irregular in nature since PV system output power is intermittent and highly dependent on environmental conditions. Irradiance, humidity, PV surface temperature, and wind speed are only a few of these variables. The uncertainty in photovoltaic generating, it's crucial to plan ahead for solar power generation. Solar power forecasting is required for electric grid supply and demand planning. Because solar power generation is weather-dependent and unregulated, this forecast is complicated and difficult. Selective developed to this goal. Traditional approaches such as statistics, autoregressive moving average, regression, and others were used to forecast PV power before the widespread usage variables are assessed for prediction models based on Artificial Neural Networks (ANN) and regression models. Several PV forecasting algorithms have been of machine learning technologies. Artificial Neural Networks, Support Vector Machines, and hybrid techniques have grown popular as a result of recent advances in machine learning methodologies and access to huge data. This study examines the impacts of numerous environmental conditions on PV system output, as well as the working principle and application of various PV forecasting approaches, in order to better comprehend the insights of PV prediction. Furthermore, the important parameters influencing PV generation are calculated using real-time data.
  • Miniatura
    Ítem
    Renewable power generation price prediction and forecasting using machine learning
    (Wiley, 2024-05-28) Rao, Challa Krishna; Sahoo, Sarat Kumar; Yanine, Fernando
    In the power market, electricity price forecasting (EPF) is a crucial consideration for decision-making. The need to optimize earnings by adjusting bids in day-ahead power markets is becoming more and more important to different market participants. Prior information is required for marketers to have an advantage over the competition while controlling the risk of pricing fluctuation. However, not all marketers must accurately predict the worth of future pricing when making decisions. To make a choice, it is necessary to determine whether the cost will be prohibitive. Thus, in order to determine the prices that have an impact on marketers, electricity price classification is first performed. Based on a threshold value, prices are categorized as low class prices and high class pricing. In order to determine the precise value of prices for utility maximization, the EPF has been explored next. In order to maximize advantage or utility, buying and selling bidding techniques rely on the accurate projections of prices for the following day. Effective forecasting models can improve the performance of producers and consumers, who play important roles in the electrical markets. The EPF approaches now in use produce complicated models and are not generalizable.The best method for forecasting prices with superior generalization performance, kernel functions, and distributive prediction is to use machine learning (ML)-based models. In order to forecast prices, machine learning-based electricity price forecasting has been studied. On the markets of Ontario, Austria, and India, the ML models have been put to the test. Neural networks, support vector machines, core vector machines, information vector machines, and relevance vector machines (RVM) have all been used with EPF, and their performances have been assessed in order to determine the optimum model. When all of these models are examined, it becomes clear that RVM is the most effective and trustworthy ML approach for various markets. The price of power fluctuates a lot, with both typical costs and spikes that might be tens to hundreds of times higher than the average price range. The prediction for the overall price of energy is created by adding together average prices and price peaks. A good and precise price spike forecast is required because price spikes have a big influence on the electrical market. As a result, a threshold value has also been used to forecast the spike value. The potential for more research in this field has also been emphasized.