研究目的
To compile knowledge about solar power forecasting, focus on recent advancements, and compare multiple regression techniques including linear least squares and support vector machines for short-term day-ahead solar radiation prediction.
研究成果
The SVM model with RBF kernel (γ=0.035) performs best for short-term solar radiation prediction, achieving a root mean square error improvement of around 29% compared to other models. This approach is feasible and effective for enhancing grid stability and cost-effective energy dispatch. Future work should focus on refining models and expanding data sources for broader applicability.
研究不足
The study relies on historical data from specific locations (e.g., Nadi Airport in Fiji), which may limit generalizability. The forecasting models may not account for all environmental variables, and the accuracy is dependent on the quality and completeness of the input data. Potential areas for optimization include incorporating more diverse datasets and advanced machine learning techniques.
1：Experimental Design and Method Selection:
The study uses machine learning techniques, specifically linear least squares regression and support vector machines (SVM) with multiple kernel functions (linear, polynomial, radial basis function) for forecasting solar radiation. The methodology involves training models on historical data to predict day-ahead solar intensity.
2：Sample Selection and Data Sources:
Historical data from a weather station and NWS weather forecasts are used, collected over 10 months from January 2012-2015. Data includes weather metrics and solar intensity in watts per m, with datasets from Nadi Airport in Fiji consisting of 1435 data samples.
3：Data includes weather metrics and solar intensity in watts per m, with datasets from Nadi Airport in Fiji consisting of 1435 data samples.
List of Experimental Equipment and Materials:
3. List of Experimental Equipment and Materials: No specific equipment or materials are mentioned beyond data sources and software (Weka-3.9.1).
4：1).
Experimental Procedures and Operational Workflow:
4. Experimental Procedures and Operational Workflow: Data is collected, preprocessed, and used to train regression models. Models are optimized, trained, and tested using Weka software. Forecast accuracy is evaluated using MAE and RMSE metrics.
5：Data Analysis Methods:
Statistical measures such as Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) are calculated to compare model performance. Kernel parameters (e.g., γ for RBF kernel) are tuned for optimal results.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
