燃气轮机稳态试验数据异常值剔除方法研究  

Research on Elimination Method of Abnormal Data in Steady-State Test of Gas Turbine

作　　者：王柏贺 WANG Baihe(Shenyang Engine Research Institute,Aero Engine(Group)Corporation of China,Shenyang 110015,China)

机构地区：[1]中国航空发动机集团有限公司沈阳发动机研究所,沈阳110015

出　　处：《热力透平》2024年第3期165-169,221,共6页Thermal Turbine

摘　　要：为确保燃气轮机整机性能评定的准确性,需对台架和机载性能测试数据进行稳态平均值求解。然而实际试验数据存在较多异常值,其对稳态平均值计算结果产生了较大的干扰。介绍了3种燃气轮机稳态试验数据异常值剔除的方法及其原理,包括莱茵达准则、均值迭代法及C模糊聚类方法,并以燃气轮机试验数据为对象,采用上述3种异常值剔除方法对试验数据进行了处理。计算结果表明:莱茵达准则、均值迭代法和C模糊聚类方法的数据处理精度基本一致,均适用于燃气轮机稳态试验数据异常值的剔除。研究成果可以为燃气轮机整机稳态试验数据处理提供参考。In order to ensure the accuracy of the performance evaluation of gas turbine,it is necessary to get the steady-state average value of the data of bench and airborne performance test.However,there is many abnormal data in actual test,which has great interference on the calculation results of steady-state average value.The principles of three data processing methods are presented,namely Rheinda criterion,mean iterative method and C fuzzy clustering method,and using the test data of a gas turbine as research objective,the abnormal data is processed by the three methods above.The results show the calculation accuracy of the three methods are generally the same,and the three are all suitable for the elimination of abnormal data value in the steady-state test of gas turbines.The research results can provide reference for data processing in steady-state test of gas turbine.

关 键 词：燃气轮机 异常值剔除方法 均值迭代法 莱茵达准则 C模糊聚类方法 

分 类 号：TK477[动力工程及工程热物理—动力机械及工程]