【摘要】 对电厂现场采集的测量数据进行不良值的检验非常必要,在对测量参数相关性分析的基础上,采用主成份分析方法进行建模,即首先对样本的高维变量数据进行标准化处理,建立相关矩阵,计算特征值和特征向量,然后求取累计方差贡献。通过数据重构实现对各测量数据正确的估计,研究了测量数据的PCA分析过程与数据结构化重构的向量投影方法,得出各个测量参数的结构化重构余差,可以对故障仪表进行正确的检测与定位。对某电厂600 MW机组汽轮机抽汽压力数据的主成分分析和检验结果表明了该方法的有效性。更多还原

【Abstract】 Checking the field data in the power plant for errors is very necessary.The model was built with the principal component analysis(PCA) method based on the analysis of correlation,i.e.,at first,the high-dimension data of the sample was processed in a standard way,the related matrix was established and the characteristic value and vector were calculated;and then the accumulative variance dedication was obtained.The measuring data was estimated accurately by data reconstruction.PCA of measuring data and vector projection methods were studied.The structured reconstruction residual of measuring parameters was calculated which can be used to accurately detect and locate the fault instrument.The effectiveness of this method was proved by PAC and the test result of the 600MW unit draw steam pressures of a domestic power plant.更多还原