摘要
光伏电站的发电功率因受不同客观环境因素的影响,其变化规律很难有迹可循,因此对光伏出力进行准确预测是实现光能大规模开发及利用的重要手段。研究将温度以及历史发电功率数据作为输入变量,提出了一种将模糊聚类(Fuzzy C-means)分析法与回声状态网络(Echo State Network)算法相结合的模型对样本进行训练和预测,并利用自适应全局和声搜索(Adaptive Global Harmony Search,AGHS)算法优化此模型,最后通过AGHS-FCM-ESN模型与传统的FCM-ESN模型进行预测误差比对,证明此模型可有效提高传统FCM-ESN模型的预测精度,并具有一定的实用性,可确保电网安全稳定地运行。
The power generation of photovoltaic power plants is affected by various environmental factorsand its chang mode is difficult to trace.Therefore,accurate prediction of photovoltaic output is an impor-tant means to realize large-scale development and utilization of solar energy.As an input variable,histori-cal power generation power data is proposed.A model combining fuzzy c-means analysis and echo statenetwork algorithm is used to train and predict samples.The Adaptive global harmony search(AGHS)al-gorithm is used to optimize this model.Finally,the prediction error is compared with the traditionalFCM-ESN model by the AGHS-FCM-ESN model,which proves that this model can effectively improvethe prediction of the traditional FCM-ESN model.Accuracy and practicality can ensure safe and stableoperation of the grid.
作者
曹青
田丽
王芳勇
CAO Qing;TIAN Li;WANG Fangyong(College of Electrical Engineering,Anhui Polytechnic University,Wuhu 241000,China;Wuhu Power Generation Co.,Ltd.,Wuhu 241000,China)
出处
《安徽工程大学学报》
CAS
2019年第1期31-35,共5页
Journal of Anhui Polytechnic University
基金
安徽高校自然科学研究重点基金资助项目(KJ2018A0121)
关键词
光伏发电功率预测
模糊聚类
回声网络算法
自适应全局和声搜索算法
photovoltaic power generation prediction
fuzzy clustering
echo network algorithm
adaptiveglobal harmony search algorithm