摘要
针对现有光伏发电功率超短期预测法,建模复杂、准确度低,难以满足实际需求的问题,提出一种综合考虑太阳辐射和光伏输出功率混沌特征提取的DMCS-WNN组合预测法。首先,在分析影响光伏发电输出功率各外界因素基础上,利用C-C法进行混沌吸引子重构,挖掘数据自身蕴含的影响发电功率的各因素信息,并根据相关性分析,选择太阳辐射作为附加输入因子;然后,针对小波神经网络(wavelet neural network,WNN)初值敏感性不足,利用布谷鸟搜索算法(cuckoo search,CS)进行寻优,并提出一种双模式布谷鸟搜索算法(dual-modecuckoosearch,DMCS)以提高传统CS的收敛速度和搜索能力;最后,建立光伏发电功率DMCSWNN预测模型,并基于实例仿真,分析其在突变和非突变天气情况下的预测效果。结果表明:该预测法在各天气类型中均保持良好的预测准确度和适用性。
In order to solve the problem of existing ultra-short-term photovoltaic(PV)power prediction methods that the modeling is complex,the accuracy is low and hardly to meet the actual requirements.A DMCS-WNN prediction method based on considering of solar radiation and chaotic feature extraction was proposed.Firstly,based on the analyze of each external factors which affect the output power of PV system,the chaotic phase space reconstruction was carried out by using C-C method to mine the hidden information of each influencing factors contained in historical PV power data,follow by this,selecting solar radiation as additional input according to the correlation analysis.Secondly,aiming at the shortcomings of wavelet neural network(WNN),its sensitivity to initial value was improved by cuckoo search(CS)algorithm,and a dual-mode cuckoo search(DMCS)algorithm was proposed to improve the convergence speed and search ability of traditional CS.Finally,a DMCS-WNN prediction model for PV power generation was established,and the prediction results under abrupt and non-abrupt weather conditions were verified based on case simulation.The results show that the proposed method maintains well prediction accuracy and applicability in each weather type.
作者
王育飞
付玉超
薛花
WANG Yufei;FU Yuchao;XUE Hua(College of Electrical Engineering,Shanghai University of Electric Power,Yangpu District,Shanghai 200090,China)
出处
《中国电机工程学报》
EI
CSCD
北大核心
2019年第S01期63-71,共9页
Proceedings of the CSEE
基金
国家自然科学基金项目(61873159)
上海市自然科学基金项目(15ZR1418000).
关键词
光伏发电
功率预测
混沌
小波神经网络
布谷鸟算法
PV power generation
power prediction
chaotic
wavelet neural network
cuckoo search algorithm
