Enhanced Multivariate Process Monitoring for (original) (raw)

Multivariate statistical methods for the analysis, monitoring and diagnosis of process operating performance are becoming more important because of the availability of on-line process computers which collect measurements on large numbers of process variables. Principal component analysis (PCA) has been used successfully as a MSPC tool for detecting faults by extracting feature information from complex data. However the traditional linear PCA is not well adapted to complicated nonlinear systems; therefore Non-linear principal component analysis (NLPCA) is a nonlinear generalization of standard linear PCA. NLPCA can be achieved by using a neural network with an autoassociative architecture. In order to monitor process performance, the SPE index is used for anomalies detection step. Then the fault localization by exploiting reconstruction method as an alternative to contribution plots approach. The obtained results on real data demonstrate the technique effectiveness applied for monitoring wastewater treatment plant.  Index Terms-process monitoring, multivariate statistical process control, diagnosis, wastewater treatment plant, auto-associative neural network, nonlinear principal component analysis