An innovative machine learning based on feed-forward artificial neural network and equilibrium optimization for...

Area of Science:

• Renewable Energy Systems
• Artificial Intelligence in Energy
• Meteorological Forecasting

Background:

• Reliable energy source analysis is vital for sustainable development.
• Solar energy offers a clean, abundant alternative to fossil fuels.
• Accurate solar irradiance forecasting is essential for efficient solar power generation.

Purpose of the Study:

• To propose novel hybrid models for enhancing solar irradiance (I_S) forecasting.
• To investigate the performance of various metaheuristic algorithms in optimizing a Feed-Forward Artificial Neural Network (FFANN) for I_S prediction.
• To identify key meteorological and temporal factors influencing solar irradiance.

Main Methods:

• A Feed-Forward Artificial Neural Network (FFANN) was employed to model the non-linear relationship between I_S and meteorological/temporal parameters.
• Metaheuristic algorithms, including Equilibrium Optimization (EO), Wind-Driven Optimization (WDO), Optics Inspired Optimization (OIO), and Social Spider Algorithm (SOSA), were used to optimize the FFANN.
• Principal Component Analysis (PCA) was applied to determine the most significant input factors.

Main Results:

• Metaheuristic algorithms successfully trained the FFANN, demonstrating high accuracy on 80% of the data.
• The trained models exhibited strong predictive capabilities on unseen data (20% of the dataset).
• Equilibrium Optimization (EO) outperformed WDO, OIO, and SOSA in achieving higher prediction accuracy.
• PCA identified key factors influencing solar irradiance, enabling potential dimension reduction and practical recommendations.

Conclusions:

• Hybrid models integrating FFANN with metaheuristic optimization provide a proficient approach for solar irradiance forecasting.
• Equilibrium Optimization is a highly effective metaheuristic for improving solar irradiance prediction accuracy.
• The study provides an explicit formula derived from the EO-based solution for convenient solar irradiance estimation and offers insights for enhancing solar energy production.