摘要
研究太阳辐射对认识气候变化有重要影响,而中国地面台站的太阳辐射观测数据不足,质量不佳。针对提高太阳辐射观测的精确度,提出利用主成分分析法(PCA)对影响太阳辐射观测的多个气象要素进行降维处理,剔除冗余变量,结合遗传算法(GA)获取BP模型的最优权值阈值,并用实测数据加以验证该模型的可行性,充分提高模型性能以实现对太阳辐射的观测研究。结果表明:基于PCA的GA-BP网络模型的观测精度高于传统BP模型,该模型有效地提高模型的泛化能力,具有一定的可行性及指导意义。
The study of solar radiation has an important impact on the understanding of climate change,while the solar radiation observation data of ground stations in China is insufficient and of poor quality. In order to improve the accuracy of solar radiation observation,this paper proposes a method of principal component analysis（ PCA） to reduce the number of meteorological factors that influence the solar radiation,which eliminate redundant variables. Combining with the genetic algorithm（ GA） to obtain the optimal weights and thresholds of BP model,and using the measured data to verify the feasibility of the model,it can fully improve the network performance to achieve the observation of solar radiation. The results show that the accuracy of GA-BP network model based on PCA is higher than traditional BP model,which can improve the generalization ability of the model and has certain feasibility and guidance significance.
出处
《成都信息工程大学学报》
2017年第6期584-589,共6页
Journal of Chengdu University of Information Technology
基金
省科技厅科技支撑资助项目(2015GZ0278)
关键词
太阳辐射
主成分分析
遗传算法
BP神经网络
误差分析
solar radiation
principal component analysis （PCA）
genetic algorithm （GA）
BP neural network （BPNN）
error analysis
