摘要
为了从光伏发电历年数据中筛选出有效样本,并提高光伏发电量预测模型的准确率,文章将聚类分析应用于光伏领域,并结合神经网络建立了光伏发电量预测模型。以晴天、多云和雨天3种天气类型为目标,利用聚类分析对历史数据中的异常样本进行筛选,并将筛选后的样本作为训练数据建立了反向传播(BP)神经网络预测模型。通过对比筛选前后预测模型的计算结果可知,利用聚类分析筛选后的数据所建立起来的预测模型精度较高,因此,聚类分析和BP神经网络相结合是提高光伏发电量预测精度的一种有效方法。
In order to select the effective samples in the data of PV power generation years and to improve the accuracy of PV power generation forecasting model,this paper uses clustering analysis for this field and establishes forecasting model based on neural network.Based on three different types of weathers on sunny,cloudy and rainy days,the anomalous samples in historical data were screened by cluster analysis,and the back propagation （BP） neural network prediction model was established using the filtered samples as training data.By comparing the results of the model before and after the data screening,we can see that the data prediction model established after cluster analysis is more accurate.Therefore,the combination of cluster analysis and BP neural network is an effective method to improve the prediction accuracy of photovoltaic power generation.
作者
成珂
郭黎明
王亚昆
Cheng Ke Guo Liming Wang Yakun(School of Power and Energy, Northwestern Polytechnical University, Xi＇an 710072, China)
出处
《可再生能源》
CAS
北大核心
2017年第5期696-701,共6页
Renewable Energy Resources
基金
陕西省科学技术研究发展计划(2015XT-07)
关键词
聚类分析
数据筛选
神经网络
光伏发电量预测
太阳能
cluster analysis
data screening
neural network
photovoltaic power generation prediction
solar energy
