基于模糊熵聚类的空气预热器性能监测

Performance Monitoring of the Air Pre-heater Based on Entropy Fuzzy Clustering

在基本模糊熵聚类方法(EFC)的基础上加入一个用于统计特征的权重因子,提出一种改进的加权熵聚类方法(W-EFC),并将其应用于电站空气预热器的堵灰监测过程。从测试数据集的聚类结果可以看出,W-EFC具有较好的离群点识别效果,并在一定程度上降低了噪声对数据的影响。继而,以空气预热器历史运行数据为研究对象,完成W-EFC聚类,聚类结果可以获得较长运行时间内不同的工况和不同的性能水平,可为实时运行监测指导提供新思路。

On the basis of the basic fuzzy entropy clustering method(EFC),a weight factor for statistical characteristics was added,and an improved weighted entropy clustering method(W-EFC) was proposed and was employed to monitor blockage of the air pre-heater in a power station.Clustering the data set to show that,the W-EFC has a good effect on outlier identification and it can reduce the noise influence on the data to a certain extent.In addition,through taking the historical operation data of the air pre-heater as the object of research,the W-EFC clustering was completed and through it,different working conditions and different performance levels in a long operation time can be obtained,which provides new ideas for real-time monitoring of the operation.