基于FCM和CG-DBN的光伏功率短期预测  

Short-term Prediction of Photovoltaic Power based on Combination of FCM and CG-DBN

作　　者：李正明[1] 高赵亮 梁彩霞 LI Zhengming;GAO Zhaoliang;LIANG Caixia(College of Electrical and Information Engineering,Jiangsu University,Zhenjiang 212013,China)

机构地区：[1]江苏大学电气信息工程学院

出　　处：《现代电力》2019年第5期62-67,共6页Modern Electric Power

基　　金：国家自然基金项目（51477070）;江苏高校优势学科建设工程资助项目PAPD（20116）

摘　　要：针对光伏输出功率非线性、波动大、不稳定等特征引起光伏功率短期预测不精确的问题,本文提出了一种基于相似日聚类和利用共轭梯度法(CG)改进深度信念网络(DBN)的组合模型预测方法。首先利用FCM聚类算法将原始数据按照隶属度进行相似日聚类,随后根据类别进行CGDBN预测模型的建模,最后利用该模型进行光伏输出功率的短期预测。本文将方案应用于浙江龙游发电站,并将预测结果与传统预测模型进行了比较。最终得出,FCM和CG-DBN组合预测模型在光伏功率短期预测中的性能优于其他模型。Aiming at nonlinearity,large fluctuation and instability of photovoltaic output power,this paper proposes a combined prediction model based on similar-day clustering and improved deep belief network(DBN)with conjugate gradient(CG).Fuzzy C-means clustering algorithm(FCM)is firstly used to perform similar-day clustering from raw data according to the membership degree.Then the CG-DBN prediction model is built based on the obtained categories to predict the short-term PV output.The scheme is applied to Longyou power station in Zhejiang province.The prediction results are compared with the traditional prediction model and prediction performance of the FCM and CG-DBN combination forecast model is better than other models.

关 键 词：相似日聚类 深度信念网络 光伏功率短期预测 组合预测模型 共轭梯度法 

分 类 号：TM732[电气工程—电力系统及自动化]