基于特征子空间的KPCA及其在故障检测与诊断中的应用

Feature subspace based KPCA and its application to fault detection and diagnosis

针对标准KPCA(kernelprincipalcomponentanalysis)不适合大样本分析的缺点,提出了一种基于特征子空间的KPCA(FSKPCA)及其故障检测与诊断方法,该方法通过构建具有较小维数的特征子空间上的正交基来简化核矩阵,从而降低KPCA的计算复杂性.与标准KPCA方法相比,FSKPCA方法具有更高的计算效率且只需较小的计算机存储空间.通过非等温连续反应釜过程的故障检测与诊断的应用实例,说明了本算法的有效性.

A feature subspace based kernel principal component analysis （KPCA）, method （FS KPCA） and its application to fault detection and diagnosis are presented in this paper to overcome the shortcoming of the standard KPCA method which is not appropriate to deal with a large number of training data. FS KPCA simplifies the kernel matrix and reduces the computational cost of KPCA by constructing a lowerdimensional orthonormal based on feature subspace. When applied to process monitoring, the FS_ KPCA- based method is more efficient in computation and needs less computer memory than standard KPCA-based methods. Computer simulation of non-isothermal CSTR process monitoring demonstrates the effectiveness and efficiency of the proposed method.