基于动态PCA的核动力装置传感器故障检测

针对变工况过程中传统主元分析方法的模型不适应问题，通过稳定性因子分析，剔除过渡过程数据，并用模糊聚类方法将不同稳态工况进行分类，利用动态主元模型方法根据工况类型建立不同的主元模型，并将该方法用于核动力装置传感器的故障检测，结果表明该方法能够适应变工况情况下的传感器故障检测，减少了故障的误检，并提高了检测灵敏度．

Sensor Fault Detection for Nuclear Power Plant Based on Dynamic Principal Component Analysis

As to the maladjustment of model of traditional principal component analysis in changing condition process, different principal component models have been built by dynamic principal compo- nent analysis according to condition type, through stability factor analysis to eliminate the changing process data and condition classification of different steady conditions with the fuzzy-clustering meth- od. This method is applied to sensor fault detection for nuclear power plant . The result shows that it is fit for sensor fault detection in changing condition process, it reduces the chances of detection mis- takes and it improves the detection sensitivity.