摘要
针对风电测量数据中不同程度存在的异常数据势必影响风电预测精度的问题,结合滑窗方法并运用两种智能算法实现了风电测量数据异常识别。首先分析了风电测量数据的时序性,然后描述了基于高密度连通区域的聚类算法和基于密度的离群点检测算法,并分析了两种方法的性能。最后,针对国内风电场实际测量数据,运用两种算法进行了异常测量数据的识别并在此基础上进行了风电预测,针对计算结果进行了比较分析,验证了所提策略的可行性和有效性。
Various degrees of abnormal data in wind power measurement are bound to affect wind power prediction accuracy. This paper combines with moving window and utilizes two intelligent algorithms to implement the identification of abnormal measurement data of the wind power. First, it analyzes time character of wind power measurement data. Then it describes density-based spatial clustering of application with noise （DBSCAN） and density-based outlier detec- tion （DLOF） , and analyzes the performance of two methods. Finally, for domestic wind farm actual measurement data, two kinds of algorithms are used to recognize abnormal measurement data. On this basis, wind power prediction is car- ried out; computation results are compared and analyzed; and the feasibility and effectiveness of the proposed strategy are verified.
出处
《仪表技术》
2017年第1期10-14,共5页
Instrumentation Technology
关键词
风电测量数据的时序性
异常测量数据识别
风电功率预测
time character of wind power measurement data
abnormal measurement identification
wind power prediction
