Adaptive Sampling for Monitoring Multi-Profile Data with Within-and-between Profile Correlation

Multi-profile data can provide within-and-between profile information for efficiently modeling and monitoring system status. In practice, however, acquisition of such requires large number sensors, which raises various concerns difficulties, example, cost, energy, transmission bandwidth, in accessing the full from each sensor. this article, we propose an adaptive sampling strategy multi-profile by using limited portion data. The proposed scheme incorporates correlation features balance between random search greedy identifying most informative profiles. More specifically, multivariate functional principal component analysis (MFPCA) is used to capture correlation, MFPC scores are augmented unobservable profiles feed into a CUSUM chart. Two properties method allocating resources among sensors investigated. Numerical case studies conducted under scenarios demonstrate effectiveness method.