Data on Support Vector Machines (SVM) model to forecast photovoltaic power

Area of Science:

• Renewable Energy Systems
• Machine Learning Applications
• Data Science

Background:

• Accurate photovoltaic (PV) power forecasting is crucial for grid integration and energy management.
• Traditional forecasting models often struggle with high-dimensional weather data and computational complexity.
• Dimensionality reduction techniques can improve the efficiency and accuracy of PV power prediction models.

Purpose of the Study:

• To develop and evaluate a hybrid forecasting model for PV power.
• To investigate the effectiveness of data dimensionality reduction in PV power forecasting.
• To predict PV power output for various time horizons (1-24 hours ahead).

Main Methods:

• Hybrid model combining Principal Component Analysis (PCA) for dimensionality reduction and Least Squares Support Vector Machines (LS-SVM) for prediction.
• Application of quadratic Renyi entropy criteria within the PCA framework.
• Forecasting PV power on an hourly basis for day-ahead predictions.

Main Results:

• The hybrid PCA-LS-SVM model demonstrates effective PV power forecasting capabilities.
• Data reduction using PCA significantly enhances model efficiency without compromising accuracy.
• Accurate hourly PV power predictions were achieved for multiple ahead hours.

Conclusions:

• The proposed hybrid approach offers a robust and efficient method for PV power forecasting.
• Dimensionality reduction is a valuable technique for optimizing complex PV forecasting models.
• The model provides reliable predictions essential for renewable energy grid management.