Processes going on in modern chemical processing plants are typically very complex,and this complexity is also present in collected data, which contain the cumulative effectof many underlying phenomena and disturbances, presenting different patterns in thetime/frequency domain. Such characteristics motivate the development and applicationof data-driven multiscale approaches to process analysis, with the ability of selectivelyanalyzing the information contained at different scales, but, even in these cases, there isa number of additional complicating features that can make the analysis not beingcompletely successful. Missing and multirate data structures are two representatives ofthe difficulties that can be found, to which we can add multiresolution data structures,among others. On the other hand, some additional requisites should be considered whenperforming such an analysis, in particular the incorporation of all available knowledgeabout data, namely data uncertainty information.In this context, this thesis addresses the problem of developing frameworks that are ableto perform the required multiscale decomposition analysis while coping with thecomplex features present in industrial data and, simultaneously, consideringmeasurement uncertainty information. These frameworks are proven to be useful inconducting data analysis in these circumstances, representing conveniently data and theassociated uncertainties at the different relevant resolution levels, being alsoinstrumental for selecting the proper scales for conducting data analysis.In line with efforts described in the last paragraph and to further explore the informationprocessed by such frameworks, the integration of uncertainty information on commonsingle-scale data analysis tasks is also addressed. We propose developments in thisregard in the fields of multivariate linear regression, multivariate statistical processcontrol and process optimization.The second part of this thesis is oriented towards the development of intrinsicallymultiscale approaches, where two such methodologies are presented in the field ofprocess monitoring, the first aiming to detect changes in the multiscale characteristics ofprofiles, while the second is focused on analysing patterns evolving in the time domain.
展开▼