Improved anomaly detection using multi-scale PLS and generalized likelihood ratio test
KAUST DepartmentComputer, Electrical and Mathematical Sciences and Engineering (CEMSE) Division
Applied Mathematics and Computational Science Program
KAUST Grant NumberOSR-2015-CRG4-2582
MetadataShow full item record
AbstractProcess monitoring has a central role in the process industry to enhance productivity, efficiency, and safety, and to avoid expensive maintenance. In this paper, a statistical approach that exploit the advantages of multiscale PLS models (MSPLS) and those of a generalized likelihood ratio (GLR) test to better detect anomalies is proposed. Specifically, to consider the multivariate and multi-scale nature of process dynamics, a MSPLS algorithm combining PLS and wavelet analysis is used as modeling framework. Then, GLR hypothesis testing is applied using the uncorrelated residuals obtained from MSPLS model to improve the anomaly detection abilities of these latent variable based fault detection methods even further. Applications to a simulated distillation column data are used to evaluate the proposed MSPLS-GLR algorithm.
CitationMadakyaru M, Harrou F, Sun Y (2016) Improved anomaly detection using multi-scale PLS and generalized likelihood ratio test. 2016 IEEE Symposium Series on Computational Intelligence (SSCI). Available: http://dx.doi.org/10.1109/SSCI.2016.7849880.
SponsorsThis publication is based upon work supported by the King Abdullah University of Science and Technology (KAUST) Office of Sponsored Research (OSR) under Award No: OSR-2015-CRG4-2582.
Conference/Event name2016 IEEE Symposium Series on Computational Intelligence, SSCI 2016