基于相似日的光伏发电短期预报模型  

Short-term Power Forecasting Model for PV Generation Based on Similar Day

作　　者：罗小兰[1] 刘士荣[1] 郑凌蔚[1] 竺健[1] 

机构地区：[1]杭州电子科技大学自动化学院,浙江杭州310081

出　　处：《杭州电子科技大学学报（自然科学版）》2015年第5期74-78,共5页Journal of Hangzhou Dianzi University：Natural Sciences

基　　金：浙江省自然科学基金资助项目(LQ12E07001);浙江省重大科技专项资助项目(2009C11020)

摘　　要：在分析影响光伏阵列发电功率的天气因素基础上,提出了一种基于天气预报信息和相似日分类选取的反馈型神经网络预测模型。运用天气预报和相似日评价函数,从天气的历史数据中选择出天气相似日,结合相似日的发电功率和相似日与预测日的天气数据建立Elman神经网络预测模型。实验结果表明,模型预测精度较好,具备一定有效性。With the analysis of factors effecting the generated power of PV arrays,a feedback neural network prediction model was proposed based on the weather forecasting information and similar-day selection method in this paper.In the presented methodology,the similar-day was picked from the historical data using the evaluation function and the weather forecasting information.In addition,the applied Elman neural network prediction model was built based on the power generation data in the similar-day and the weather data of similar-day and the predicted date.The experiments show that the forecasting model has high prediction accuracy.

关 键 词：光伏发电 功率预报 ELMAN神经网络 相似日分类 

分 类 号：TM615[电气工程—电力系统及自动化]