基于数据挖掘和模糊聚类的风电功率实时预测研究  

Real-time prediction of wind power based on data mining and fuzzy clustering

作　　者：杨茂[1] 熊昊[1] 严干贵[1] 穆钢[1] 

机构地区：[1]东北电力大学微通电力研究室,吉林吉林132012

出　　处：《电力系统保护与控制》2013年第1期1-6,共6页Power System Protection and Control

基　　金：国家自然科学基金重点项目(60934005);吉林省产业技术研究与开发专项项目(JF2012C018)~~

摘　　要：准确的风电功率预测是实现风能大规模开发利用的有效手段,实时预测能够滚动地修正日出力计划曲线,保证电力系统运行的安全性和经济性。在分析风电场不同机组出力特性的基础上,利用数据挖掘和模糊聚类技术将不同的机组进行分类,并分别进行实时预测,将预测结果进行累加得到最终的预测结果。以中国吉林省某风电场的实测风电数据为例,进行了实时预测,结合国家能源局对风电功率实时预测预报管理要求中的指标进行了分析。结果表明,所提出的方法准确率最大提高2.57%,合格率最大提高4.23%,均方根误差最大下降3.21%,说明了该方法的有效性。Accurate real-time prediction of wind power is an effective means to achieve rational management of wind energy in large-scale. Real-time prediction can modify the day-output planned curve, then ensure the security and economy of power system. The characteristics of different unit output of the wind farm are analyzed, data mining and fuzzy clustering techniques are adopted to classify the different units of wind farm, real-time prediction is performed for each class, and the predicted results are cumulative for the final prediction results. Through the measured data of wind farms in Jilin Province, we perform the real-time prediction and make an analysis combining with the index in the management requirements about wind power real-time prediction from the National Energy Board. The results show that the precision rate with the proposed method rises by 2.57% at maximum, the qualified rate rises by 4.23% at maximum, and the root mean square error declines by 3.21% at maximum, indicating the effectiveness of this method. This work is supported by National Natural Science Foundation of China （No. 60934005）.

关 键 词：风电功率 实时预测 数据挖掘 模糊C均值聚类 

分 类 号：TM614[电气工程—电力系统及自动化] TM73