基于Fisher有序聚类的汽轮机试验数据稳态检测方法

Steady-state Detection of Turbine Performance Test Data Based on Fisher Ordered Clustering Method

热力试验数据中包括"稳态数据"和"暂态数据",其中稳态数据的检测是完成汽轮机性能试验数据计算分析的前提,目前主要以人工选择的方式确定热力试验稳态数据。提出基于Fisher有序聚类方法对汽轮机性能试验数据分段处理,首先对汽轮机性能试验采集数据进行首次分割,通过稳定度分析判断Fisher划分后的数据段的相对稳定性,其次对不满足稳定度要求的已分割数据段进行再分割,以提高稳态数据检测精度,最终提取出符合汽轮机性能试验计算和分析要求的稳态数据。

Detection of steady-state data is a must operate after data collection in steam turbine performance test. The use of Fisher ordered clustering is proposed to divide the collect data of turbine performance test for the first time. Then determine the relative stability of the divided data after Fisher ordered clustering by stability analysis, and the segmented data will be subdivided if does not meet the stability requirements. We can extract steady state data that meet the requirements of turbine performance test calculation and analysis eventually.