研究目的
To forecast the generated power of photovoltaic (PV) power station using general regression (GR) and back propagation (BP) neural network prediction methods, and to compare the effectiveness of these two models.
研究成果
The BP neural network prediction method was found to have better prediction results than the GR method for PV power generation. Both models are significant for power system scheduling and coordinated operation, with BP neural network being more suitable for accurate predictive control and GR method for large data groups.
研究不足
The study does not mention specific technical constraints or areas for optimization in the experimental setup or data analysis methods.
1：Experimental Design and Method Selection:
The study utilized Pearson correlation coefficient method to analyze meteorological factors and selected irradiance and battery temperature as important influencing variables. Weather was classified into sunny, cloudy, and rainy days using learning vector quantization (LVQ) neural network. GR and BP neural network prediction methods were then applied for short-term prediction of PV power system.
2：Sample Selection and Data Sources:
Historical data of solar irradiance, air temperature, and power output between May 2016 and May 2015 with a 5-min interval of a 15 kW PV system in Ashland were used.
3：List of Experimental Equipment and Materials:
Not explicitly mentioned.
4：Experimental Procedures and Operational Workflow:
The process involved correlation analysis, weather classification, and model establishment using GR and BP neural networks.
5：Data Analysis Methods:
The prediction results of both models were compared and analyzed for accuracy and fitting degree.
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
