Enhanced Anomaly Detection Via PLS Regression Models and Information Entropy Theory

Accurate and effective fault detection and diagnosis of modern engineering systems is crucial for ensuring reliability, safety and maintaining the desired product quality. In this work, we propose an innovative method for detecting small faults in the highly correlated multivariate data. The developed method utilizes partial least square (PLS) method as a modelling framework, and the symmetrized Kullback-Leibler divergence (KLD) as a monitoring index, where it is used to quantify the dissimilarity between probability distributions of current PLS-based residual and reference one obtained using fault-free data. The performance of the PLS-based KLD fault detection algorithm is illustrated and compared to the conventional PLS-based fault detection methods. Using synthetic data, we have demonstrated the greater sensitivity and effectiveness of the developed method over the conventional methods, especially when data are highly correlated and small faults are of interest.

[1]  Richard D. Braatz,et al.  Fault Detection and Diagnosis in Industrial Systems , 2001 .

[2]  Michèle Basseville,et al.  Detection of abrupt changes: theory and application , 1993 .

[3]  Fred Spiring,et al.  Introduction to Statistical Quality Control , 2007, Technometrics.

[4]  J. E. Jackson,et al.  Control Procedures for Residuals Associated With Principal Component Analysis , 1979 .

[5]  Rolf Isermann,et al.  Model-based fault-detection and diagnosis - status and applications , 2004, Annu. Rev. Control..

[6]  Jack Y. Yang,et al.  Partial Least Squares Based Dimension Reduction with Gene Selection for Tumor Classification , 2007, 2007 IEEE 7th International Symposium on BioInformatics and BioEngineering.

[7]  S. Wold,et al.  The Collinearity Problem in Linear Regression. The Partial Least Squares (PLS) Approach to Generalized Inverses , 1984 .

[8]  Hazem Nounou,et al.  PLS-based EWMA fault detection strategy for process monitoring , 2015 .

[9]  Riccardo Muradore,et al.  A PLS-Based Statistical Approach for Fault Detection and Isolation of Robotic Manipulators , 2012, IEEE Transactions on Industrial Electronics.

[10]  Raghunathan Rengaswamy,et al.  A review of process fault detection and diagnosis: Part III: Process history based methods , 2003, Comput. Chem. Eng..

[11]  Fouzi Harrou,et al.  Anomaly detection/detectability for a linear model with a bounded nuisance parameter , 2014, Annu. Rev. Control..

[12]  Klaus J. Miescke,et al.  Statistical decision theory : estimation, testing, and selection , 2008 .

[13]  Farid Kadri,et al.  Improved principal component analysis for anomaly detection: Application to an emergency department , 2015, Comput. Ind. Eng..

[14]  H. Hotelling.  Analysis of a complex of statistical variables into principal components. , 1933 .

[15]  B. Kowalski,et al.  Partial least-squares regression: a tutorial , 1986 .

[16]  Elaine B. Martin,et al.  Model selection for partial least squares regression , 2002 .

[17]  Hazem Nounou,et al.  Statistical fault detection using PCA-based GLR hypothesis testing , 2013 .

[18]  Theodora Kourti,et al.  Process analysis, monitoring and diagnosis, using multivariate projection methods , 1995 .

[19]  Fouzi Harrou,et al.  Monitoring linear antenna arrays using an exponentially weighted moving average-based fault detection scheme , 2014 .

[20]  R. A. Leibler,et al.  On Information and Sufficiency , 1951 .