In modern industrial processes, massive data are gathered to monitor and analyse a process. Intelligent analysis helps to reveal as much information of the process as possible. This information is most useful, when it is available in the form of interpretable, predictive models. Machine learning is about extracting such models from data automatically. And our, now, grid-enabled, machine learning framework (mlf) provides a suite of ml methods, which create models that are interpretable AND computational, following the black-box and white-box principles.
Paper production is a complex flow process with many process steps and levels and hundreds of process parameters influencing dozens of quality features. Paper production is so complex that the creation of explicit models is hardly possible. Intelligent data analysis, by means of ml is the right way to overcome this barrier.
For the Technology Center of VOITH Paper, a market-leading provider of large paper production systems, who built a project consortium with SCA, we conducted a "PaperMiner" project.
PaperMiner combines a GUI type fromt-end and an expert front-end. The expert front-end enables the users to configure ml tasks. This is very intuitive through mlf's Mathematica front-end.
The greatest concrete result of the PaperMiner project. SCA set a new generation paper machine, from VOITH, in operation and introduced a new high-quality Graphics paper. Due to better insight, derived from the PaperMiner support, the President of SCA SC Laakirchen could summarize, "This was an unexpected short start-up", his appreciation.