Integrated support vector regression and an improved particle swarm optimization-based model for solar radiation...

Area of Science:

• Renewable Energy
• Computational Intelligence
• Environmental Science

Background:

• Accurate solar radiation estimation is crucial for renewable energy decision-making.
• Traditional Support Vector Regression (SVR) parameter selection often relies on trial-and-error, leading to suboptimal accuracy.
• Optimization algorithms are needed to enhance SVR model performance for solar energy prediction.

Purpose of the Study:

• To develop and evaluate a novel solar radiation prediction model.
• To improve the accuracy of solar radiation estimation by optimizing SVR parameters.
• To compare the proposed model's performance against various established prediction methods.

Main Methods:

• The study integrates Support Vector Regression (SVR) with the Improved Particle Swarm Optimization (IPSO) algorithm.
• The IPSO algorithm enhances the global search capability for optimizing SVR parameters.
• The model was tested using solar radiation data from Adana, Antakya, and Konya, Turkey, and compared with M5 Tree, Genetic Programming, SVR-PSO, SVR-GA, SVR-FFA, and MARS.

Main Results:

• The proposed SVR-IPSO model demonstrated superior performance in solar radiation prediction.
• Sensitivity analysis was conducted to identify optimal input parameters for enhanced prediction accuracy.
• Performance was evaluated using multiple standard indices, confirming the model's efficiency.

Conclusions:

• The SVR-IPSO model offers a significant improvement for solar radiation prediction accuracy.
• Optimizing SVR parameters with advanced algorithms like IPSO is effective for renewable energy applications.
• The findings support the use of SVR-IPSO for reliable solar energy resource assessment.