天气分类和BP神经网络的光伏短期出力预测  

Short-term photovoltaic power output prediction method based on fuzzy weather classification and improved BP network

作　　者：蒋小波 徐小艳 刘乐平[1] 杨忠明[1] JIANG Xiao-bo;XU Xiao-yan;LIU Le-ping;YANG Zhong-ming(Guangdong Polytechnic of Science and Technology,Zhuhai Guangdong 519090,China)

机构地区：[1]广东科学技术职业学院,广东珠海519090

出　　处：《电源技术》2020年第12期1809-1813,共5页Chinese Journal of Power Sources

基　　金：广东省重大领域研发计划项目(2018B010-109003)。

摘　　要：光伏发电系统输出功率受天气因素影响,呈现随机性、波动性和间歇性。并网时可能影响电网稳定可靠运行。对此构建一个基于天气分类和改进反向传播(BP)神经网络算法的光伏出力预测模型。模型采用地表地外辐照强度相关系数、波动系数和波形熵三维特征,对33种天气类型进行量化,提取特征向量,用K-MEANS算法对特征向量进行自适应聚类,得到兼顾模型复杂度和泛化能力的4种天气类型;用遗传模拟退火(GSA)算法对BP神经网络模型的权值和阈值进行全局寻优,确保其收敛于全局最优解。依据实际光伏发电系统发电历史数据,用上述方法做出力预测,结果表明该方法预测数据与各种天气类型实际发电数据基本一致,表明该方法具有一定的推广应用价值。The output power of photovoltaic power generation system is affected by weather,showing randomness,volatility and intermittence.Grid connection may affect the stable and reliable operation of power grid.To solve this problem,a photovoltaic output prediction model based on weather classification and improved back propagation(BP)neural network algorithm was constructed.Three dimensional characteristics of correlation coefficient,wave coefficient and waveform entropy of surface radiation intensity were used to quantify 33 weather types,and feature vectors were extracted.The feature vectors were clustered adaptively by K-means algorithm to obtain four weather types taking into account the complexity and generalization ability of the model,and GSA algorithm optimized the weights and thresholds of BP neural network model globally to ensure its convergence to the global optimal solution.According to the historical data of actual photovoltaic power generation system,the above method was used to predict the output.The results show that the prediction data of this method are basically consistent with the actual power generation data of various weather types,which indicates that the method has certain popularization and application value.

关 键 词：光伏发电 短期出力预测 模糊天气分类 自适应聚类 BP神经网络 

分 类 号：TM615[电气工程—电力系统及自动化]