Abstract Recently monitoring techniques using canonical variate analysis (CVA) based state space model are proposed by several researchers. They discussed that canonical variate analysis based approach is superior to dynamic principal component analysis (PCA) based that for fault detection and identification. To apply canonical variate analysis base approach to time-varying systems, we proposed adaptive monitoring method via CVA based state space model updating. Forgetting factor is employed, to update current mean vector and correlation matrix and current state space model is recursively, estimated using Cholesky factor updating scheme. Two state space model based monitoring indices are proposed to detect process abnormalities. One is state monitoring index and the other noise process monitoring index. They are formulated using state prediction matrix and noise extraction matrix which are derived from current state space model and the statistics of them are also recursively determined using the η 2 distribution. To adjust forgetting factors according to variation of process state, the forgetting factor updating criterions are introduced. The proposed algorithm is applied to the continuous stirred tank reactor under operation condition change. Application results provide the expectation that the proposed algorithm can be applied to practical time-varying or transient processes.
[1]
S. Wold.
Exponentially weighted moving principal components analysis and projections to latent structures
,
1994
.
[2]
H. Akaike.
Markovian Representation of Stochastic Processes and Its Application to the Analysis of Autoregressive Moving Average Processes
,
1974
.
[3]
Robert J. Plemmons,et al.
Least squares modifications with inverse factorizations: Parallel implications
,
1989
.
[4]
Tsuhan Chen,et al.
Eigenspace updating for non-stationary process and its application to face recognition
,
2003,
Pattern Recognit..
[5]
In-Beum Lee,et al.
Adaptive multivariate statistical process control for monitoring time-varying processes
,
2006
.
[6]
C. Yoo,et al.
Overall statistical monitoring of static and dynamic patterns
,
2003
.
[7]
Ralph R. Martin,et al.
Merging and Splitting Eigenspace Models
,
2000,
IEEE Trans. Pattern Anal. Mach. Intell..
[8]
Masanao Aoki,et al.
State Space Modeling of Time Series
,
1987
.