IOT based solar energy prophecy using RNN architecture

It is the 21 st century and scientists say that by the end of this century, resources will be replenished and the only way the future generations can access energy is through renewable resources— those which are inexhaustible. One such source is sunlight, which has a guaranteed stay in the long run....

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:E3S web of conferences 2020-01, Vol.184, p.1007
Hauptverfasser: Vijaya Rama Raju, V., Ganapathi Raju, N V, Shailaja, V, Padullaparti, Sugandha
Format: Artikel
Sprache:eng
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:It is the 21 st century and scientists say that by the end of this century, resources will be replenished and the only way the future generations can access energy is through renewable resources— those which are inexhaustible. One such source is sunlight, which has a guaranteed stay in the long run. The energy thus given is termed as solar energy. In the present paper it is tried to solve the issue of limited resources and their adverse effects. Since the power generated from solar energy systems is highly variable, due to its dependence on meteorological conditions, an efficient method of usage of this fluctuating but precious energy source has to come in picture. This requires the scope of reliable forecast information as the development of predictive control algorithms for efficient energy management and monitoring for residential grid connected photovoltaic systems. The paper has given an overview of different applications and models for solar irradiance and photovoltaic power prediction, including time series models based on live measured data from rooftop solar power plant located at 17.5203° N, 78.3674° E. For experimentation, data collected over four years from the solar power plant was used in order to the train machine and understand the characteristics of the solar power plant and gives the predicted energy as the result. The use of Deep Learning is done where LSTM is used for the training and keras and tensorflow are used for obtaining the result. The mean square error thus obtained is 0.015.
ISSN:2267-1242
2267-1242
DOI:10.1051/e3sconf/202018401007