摘要
为了提高光伏电站辐照强度的预测精度,本文提出了基于蚁群改进BP神经网络的预测方法。首先,分析了辐照强度的影响因素,从中筛选出纬度、海拔、天气类型、日照时数、温度、空气质量、相对湿度、风速、大气压强等最优影响因子作为模型的输入;其次,通过建立新的传递函数,采用最小均方误差能量函数法进行自动优化隐含层数;按月份建立蚁群改进BP神经网络模型,对辐照强度进行预测。预测结果与BP神经网络模型进行对比,表明该方法有效提高了辐照强度的预测精度。
In order to improve the prediction accuracy of radiation intensity in photovoltaic power plants,this paper proposes a predicted method based on ant colony BP neural network. Firstly,the influencing factors are analyzed,and the optimal factors are selected as the inputs of the model,including latitude,altitude,weather type,sunshine hours,temperature,air quality,relative humidity,wind speed,atmospheric pressure. Secondly,a new transfer function is established,and the number of hidden layers is automatically optimized by the minimum mean square error energy function.At last,twelve ant colony BP network models are established by month to predict the radiation intensity. According tothe comparison of predicted results between the proposed method and BP neural network model,it is proved that theprediction accuracy is effectively improved.
出处
《电力系统及其自动化学报》
CSCD
北大核心
2016年第7期26-31,共6页
Proceedings of the CSU-EPSA
基金
国家自然科学基金资助项目(51377016)
长江学者和创新团队发展计划资助项目(IRT1114)
吉林省科技发展计划资助项目(20140101080JC)
关键词
光伏电站
辐照强度
蚁群算法
改进BP神经网络
预测
photovoltaic power plant
radiation intensity
ant colony algorithm
improved BP neural network
prediction