Machine condition monitoring by nonlinear feature fusion based on kernel principal component analysis with genetic algorithm

Feature fusion can effectively utilize complementary information from different signal sources to improve the robustness of feature extractor. As most running statuses of machines are nonlinear and non-stationary, it is difficult to extract the effective features for fault diagnosis by linear feature extractor such as PCA. Therefore, a nonlinear feature fusion scheme based on kernel principal component analysis (kernel PCA) with genetic algorithm (GA) is proposed to recognize the different conditions of rolling bearing. Kernel PCA is applied to extract higher order information from a union-vector set, in which statistical features from acoustic signals and vibration signals are incorporated. The computational problem induced by the tremendous size of the feature space is also effectively settled by using a kernel function. For better classification performance, GA is applied to search the optimal parameter in kernel function. The analytical results show that the proposed feature fusion scheme can effectively improve the recognition ability of feature extractor.