Explore chapters and articles related to this topic
Systematic Comparison of Feature Selection Methods for Solar Energy Forecasting
Published in Mohamed Lahby, Utku Kose, Akash Kumar Bhoi, Explainable Artificial Intelligence for Smart Cities, 2021
FS has been a productive scope of research and development since the seventies. It has played an important role in various fields such as data mining, machine learning, bioinformatics, and natural disaster management (Blum & Pat, 1997; Awada et al., 2012; Drotar et al., 2015; Khalid et al., 2014; Liu & Yu, 2005; Rangarajan & Veerabhadrappa, 2010). Solar energy forecasting is one of the recent and dynamic sectors, which is considered an important field of application for FS techniques. A wide range of alternative approaches has been proposed in this sense. Namely, Martin et al. (2016) have proposed a model based on FS algorithms such as linear correlation, ReliefF, and logical information analysis to improve the prediction process of solar energy production in different grid stations by selecting the most relevant meteorological attributes. Linear correlation and ReliefF methods are also addressed in (Goswami et al., 2018); authors have used the methods to select only the significant features to enhance the performance of a numerical weather prediction model. Besides, they have introduced a new FS technique that relies on local information analysis. A work elaborated in (O’Leary & Kubby, 2017) consists of using correlation-based FS to improve the artificial neural network prediction accuracy; they have shown the importance of removing noisy and complex weather features in improving the solar energy forecasting results.
Artificial neural network models for global solar energy and photovoltaic power forecasting over India
Published in Energy Sources, Part A: Recovery, Utilization, and Environmental Effects, 2020
Gulnar Perveen, M. Rizwan, Nidhi Goel, Priyanka Anand
Global warming, energy crisis, and energy security is becoming the major challenge soon and therefore it becomes necessary to use and develop alternative, sustainable, and clean energy resources for power generation. Thus, the use of Renewable Energy Sources (RES) is invigorated; since solar energy is inexhaustible in nature, therefore, considered to be the most promising renewable resources for large scale power generation. Further, the power generation from solar PV system has introduced significant economic and environmental interests and may be able to reduce the carbon-dioxide (CO2) emissions significantly and becoming a source of creating employment as well. Therefore, an accurate analysis for predicting RES can be very helpful in this perspective. For monitoring these RES, measuring equipment are installed at the projected sites; however, such meteorological sites are insufficient for providing radiation data because of costly equipment’s, its maintenance, proper calibration of the instruments, and sufficient record period. Due to unavailability of data, a need has arisen for introducing intelligent approaches for global solar energy forecasting. Further, concept of PV power generation becomes imperative for investors to market and make a decisive policy for proper energy planning. Since the solar energy changes throughout the day from dawn to dusk with changes in seasonal variation, time, and geographical locations; therefore, if investors could understand the concept of short-term photovoltaic power generation, it would help them to strategize the market policy for investment purpose.