MONITORING

Large Scale Processes

Complex systems generate large quantities of data that are being collected with the expectation of yielding useful information about the state of process operations, and lead to substantial improvements in the way they are conducted at different levels, such as: product/service quality, efficiency (reduced resources waste), flexibility (better tracking of market trends), safety and pollution prevention. However, despite the significant efforts that have been undertaken to use data more efficiently, there is still a lack of adequate approaches for handling some important features that constitute the bare nature of systems that we are currently dealing with. In short terms, these systems are inherently multivariate, multiscale, dynamic, time-varying, stochastic and are organized in complex interaction networks, that exhibit rather notable properties, such as functional hierarchy and modularity. Clearly, developing a framework for handling all these features simultaneously is still a rather ambitious objective to set, but a research path has been followed in our research group that systematically addresses proper ways of integrating several of such data characteristics.

WE DEVELOP ADVANCED MONITORING FRAMEWORKS FOR:

Continuous processes monitoring e.g., multiscale and multivariate statistical process control. Monitoring processes with multiresolution data. Dynamic PCA based on decorrelated residuals.

Profiles monitoring e.g., wavelet based profiles monitoring of 1D (paper waviness and roughness), 2D (paper formations) and 3D (cereal flakes classification and monitoring) profiles.

Batch processes monitoring e.g.,extensive comparison study of 2-way and 3-way approaches. Development of improved control limits for the monitoring statistics.