Sensor fault detection using a new online reduced rank kernel PCA for monitoring an air quality monitoring network

Process monitoring has a Great interest in the industry, due to their capacity to ensure safety operation and to maintain product quality. Therefore, the idea of this paper is to improve the fault detection performance of conventional Kernel Principal Components. In this context, a new online method based on the Reduced Rank KPCA approach has been developed for sensor fault detection of a dynamic nonlinear process. To demonstrate the efficiency of the proposed method with adaptive model compared to the conventional KPCA and the Reduced Rank KPCA, the fault detection performances are illustrated through a simulated air quality-monitoring network AIRLOR.