Authors: A. Shobana Devi; G. Maragatham; K. Boopathi; M.R. Prabu
Addresses: Department of Data Science and Business Systems, School of Computing, SRM Institute of Science and Technology, SRM Nagar, Kattankulathur – 603 203, Chengalpattu District, Tamil Nadu, India ' SRM Institute of Science and Technology, SRM Nagar, Kattankulathur – 603 203, Chengalpattu District, Tamil Nadu, India ' National Institute of Wind Energy, 657, 1A2, Velachery – Tambaram Main Rd, Pallikaranai Marshland, Pallikaranai, Chennai, Tamil Nadu 600100, India ' Indektra Power Solutions Pvt. Ltd., Plot No.8, Nagavalliamman Koil Street, Maraimalai Nagar Industrial Area, Maraimalai Nagar, Tamil Nadu 603209, India
Abstract: Solar irradiance forecasting will turn into a major challenge in the future integration of solar energy resources into existing structures of energy supply. There are squeezing requirements for approaches to accurately estimate the movement of the cloud that legitimately impacts solar power output stability. As a degree of cloudiness is concerned, cloud index images are calculated from the satellite images to derive radiation from satellite data. Solar power is predicted using a forecasting model from the predicted global horizontal irradiance (GHI). This paper focuses on forecasting the solar power of every one and half an hour using the long short-term memory (LSTM) technique. The forecasting results are compared with the actual measured 250 MW solar plant power. The experimental findings considerably increase the assessment quality of cloud movement within 15 to 90 mins that is satisfactory for grid operators to make necessary action to improve the unpredictability of solar power.
Keywords: clear sky models; cloud index; solar irradiance; satellite images; solar power forecasting.
International Journal of Powertrains, 2021 Vol.10 No.2, pp.125 - 142
Received: 29 Jun 2020
Accepted: 10 Sep 2020
Published online: 07 Sep 2021 *