Contributo alla Formalizzazione di Costrutti di Modellazione Invarianti basati sui Dati per Sistemi Cyber-Fisici

The thesis aims to identify an approach to formalize data-driven invariant modelling constructs for improving the smartness of manufacturing processes and products, involving networked components. The idea behind data-driven invariant modelling constructs is to permit the re-use of predefined functional patterns for building digital models based on the specific application. The approach makes shared knowledge more easily reusable and it is the basis of some standardization efforts. The use also of Multi Relational Data Mining techniques, in the specific case of Relational Concept Analysis (Valtchev, Missaoui, and Godin 2004), allows the extraction of tacit knowledge embedded in the (big) data coming from the analysed processes. The thesis proposes a series of modelling patterns (data-driven invariant modelling constructs) for the digital transformation of industrial production systems. A prototype for the analysis of a real industrial process on a production line at Master Italy s.r.l has been developed to experiment our knowledge extraction approach. The resulting tool can exploit existing knowledge and information from real systems to identify problems and to propose potential improvements.