Fault feature extraction of ground-testing bed based on SNGA optimized KPCA

To implement effective fault diagnosis of liquid propellant rocket engine ground-testing bed, the fault feature extraction method based on SNGA (Sharing function based niche genetic algorithm) optimized KPCA (Kernel principal component analysis) is studied. In the SNGA approach, the feature evaluation method is used to evaluate the classification ability of the feature extracted by KPCA, the comprehensive evaluation factor of all features is used as the fitness value. The hamming distance between every two individuals is defined as the sharing function. The method is experimented with the simulation data of the ground-testing bed. The comparison results of with and without optimization indicate that the feature extraction performance of KPCA is improved significantly with SNGA optimization. The comparison results of feature extraction by KPCA and PCA indicate that both feature extraction reliability and efficiency of KPCA are better than PCA in fault diagnosis of ground-testing bed.