A Multivariate Synthetic Control Chart for Process Dispersion

On-line process monitoring using local measures of association: Part I — Detection performance

“…However, the majority of the proposed multivariate statistical process control schemes are focused on the detection of shifts in the process mean [2][3][4][5], as for instance the works of Hotelling [6] and its extension to latent variable models in order to deal with the colinearity problem [7,8]. Other methods, based on multivariate cumulative sum (MCUSUM) [9][10][11][12] or multivariate exponentially weighted moving average (MEWMA) [13], are also mostly centered on the detection of deviations in the population mean vector.…”

Non-causal data-driven monitoring of the process correlation structure: A comparison study with new methods

“…A variety of multivariate statistical process control (MSPC) methods, namely control charts, have been developed and applied in order to determine whether the process is only subject to common causes of variability or if a special or assignable cause, related with some abnormality inside or outside the process, has occurred. Analysing the literature, one can verify that most multivariate process monitoring methodologies developed so far, including the latent variables methodologies (Jackson, 1959;Jackson and Mudholkar, 1979;Ku et al, 1995;Li et al, 2000;MacGregor et al, 1994;Wise and Gallagher, 1996) and state-space or time-series approaches (Negiz and Ç inar, 1997a,b), are essentially non-causal and focused on detecting changes in the process mean (Abbasi et al, 2009;Ghute and Shirke, 2008;Yeh et al, 2006;Yen et al, 2012). The important complementary problem of monitoring the process correlation structure has been almost absent from the research efforts, creating a significant gap in what regards to the high level of performance achievable today in detecting changes in the mean levels of the process variables, contrasting with the rather limited ability to effectively signal out perturbations in the correlation structure.…”

Test of covariance changes without a large sample and its application to fault detection and classification

“…To tackle this problem, one can alternatively apply the likelihood ratio test (LRT) which has a single test statistic only and possesses certain optimality property in detecting changes of parameters. Many other approaches have been also proposed to detecting change of covariance matrix, including GuerreroCusumano [1], Aparisi et al [2], Chan and Zhang [3], Levinson et al [4], Yeh and Lin [5], Yeh et al [6], Ghute and Shirke [7], Bodnar et al [8], among others. A detailed review of existing methods for detecting change of covariance matrix can also be found in Yeh et al [9].…”