Kernel locally linear embedding algorithm for quality control

Abstract

In this paper, we introduce a new multivariate statistical process control chart for outliers detection using kernel local linear embedding algorithm. The proposed control chart is effective in the detection of outliers, and its control limits are derived from the eigen-analysis of the kernel matrix in the Hilbert feature space. Our experimental results show the much improved performance of the proposed control chart in comparison with existing multivariate monitoring and controlling charts. © Springer Science+Business Media B.V. 2008.