基于核主元分析的传感器故障检测

Detection of Sensor Faults by Kernel Principal Component Analysis

提出了一种新的火电机组传感器故障检测系统。传统的主元分析方法在非线性系统中不能很好的发挥作用。采用核主元分析方法提取系统的非线性冗余信息,建立核主元模型。并在输入空间对数据进行重构,通过最小化均方预测误差来选择合适的核函数和参数,对模型的建立过程进行指导。在线检测时,利用核主元模型,将实时数据投影到核主元空间,能够有效的去除系统的噪声。对重构残差采用序贯概率比检验方法进行检验,不仅能够诊断出传感器的漂移等明显故障,而且能够及时发现设备或者系统的早期故障。通过某电厂125MW机组真空系统的多传感器故障检测仿真实例,验证了该方法的有效性。

A novel method for detecting sensor faults is being presented. The conventional principal component analysis method dosen＇t work well with non-linear systems. A model based on kernel PCA is therefore constructed, for extracting the system＇s non-linear redundant information, and then reconstructing the data in the input space. Proper kernel functions and parameters, to serve as a guide for constructing the model, are moreover chosen by way of minimizing the mean square prediction error. The model may also be effectively used during on-line fault detection service to denoise the system by projecting real-time data into the KPCA space. The sequential probability ratio test is used to detect the reconstituted residual error. Wherewith not only obvious sensor faults, like drifting, can be diagnosed, but early fault symptoms may also be intime noticed. The method＇ s effectiveness has been vindicated by a simulated example with manysided sensor faults occurring to the vacuum system of a certain 125 MW power generating set.