Horizontal Global Solar Irradiance Prediction Using Genetic Algorithm and LSTM Methods

Global horizontal solar irradiance prediction improves the capacity to manage a solar-based renewable energy system. This effective management makes the system more resilient. This could help increase its adoption, thus contributing to the reduction of greenhouse gases. Therefore, we present in this...

Full description

Saved in:
Bibliographic Details
Main Authors: Legrene, Inoussa, Wong, Tony, Dessaint, Louis-A.
Format: Conference Proceeding
Language:English
Subjects:
Forecasting
genetic algorithm
long short-term memory
Metaheuristics
Predictive models
Renewable energy sources
Solar irradiance
solar irradiance prediction
Solar panels
Windows
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Global horizontal solar irradiance prediction improves the capacity to manage a solar-based renewable energy system. This effective management makes the system more resilient. This could help increase its adoption, thus contributing to the reduction of greenhouse gases. Therefore, we present in this paper an approach to the design of a Long Short-Term Memory forecasting model using evolutionary algorithms, such as the genetic algorithm. Evaluations on different time windows have yielded interesting results. Our approach, evaluated at 6 to 72 hours, achieves Root Mean Square Error results between \mathbf{54.40} W/m^{2} and \mathbf{310.06}\ W/m^{2} and Mean Absolute Error between \mathbf{33.27}\ W/m^{2} and \mathbf{264.11}\ W/m^{2} .
ISSN:2158-2297
DOI:10.1109/ICIEA61579.2024.10665041