研究目的
To propose an adaptive solar power forecasting method that improves prediction accuracy by capturing and revising forecasting errors through data clustering, variable selection, and neural networks.
研究成果
The proposed ASPF method significantly improves solar power forecasting accuracy by adaptively revising predictions using machine learning techniques, and it enhances over time with more data, making it suitable for various forecasting models.
研究不足
The method's performance depends on the quality and quantity of historical data; it may not handle extremely complex weather changes effectively, and computational time could be a constraint for real-time applications.
1：Experimental Design and Method Selection:
The study uses an adaptive framework combining improved k-means clustering, least angular regression (LARS), and back-propagation neural network (BPNN) to revise solar power forecasts.
2：Sample Selection and Data Sources:
Solar power data from a small power system in Hekou town, Nantong city, China, collected from January to December 2016, with time intervals of 1 hour.
3：List of Experimental Equipment and Materials:
MATLAB software for simulations; no specific hardware mentioned.
4：Experimental Procedures and Operational Workflow:
Data is clustered using improved k-means, key variables are selected via LARS, BPNN is trained for error compensation, and adaptive thresholds are updated based on historical data.
5：Data Analysis Methods:
Performance evaluated using Maximum Absolute Error (MxAE), Root Mean Squared Error (RMSE), and Mean Absolute Percentage Error (MAPE).
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
