Effective energy treatment is one of the challenges to which scientists are devoting more interest and resources. Currently, there is a need to optimize the generation of energy by renewable methods and to improve the storage and management of battery cells. One of these emerging lines aims to face the problem of uncertainty that exists in the generation of energy in photovoltaic installations. In this paper, we proposed to use machine learning methods to predict the state (voltage) of the batteries at several days into the future. The results obtained from two recurrent neural networks such as NARX and LSTM are compared, getting good results with both. Two approaches are considered: sample-based prediction and pattern-based forecasting.
Battery state prediction in photovoltaic standalone installations
2020-09-10
970122 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Remote photovoltaic, standalone systems
IEEE | 1991
|Hybridization of Solar and Battery in Standalone
IEEE | 2023
|Residential and Institutional Photovoltaic Installations
AIAA | 2014
|