Knowledge Integration in Neural networks for e-Maintenance

The use of data-driven methods offers great potential in terms of improving the maintenance policies of mechanical systems. However, these methods are not without limitations and are often applied without adequately exploiting the knowledge available in the field. KINEMA aims to use state-of-the-art Machine Learning methods in an innovative way, in order to maximise the exploitation of available knowledge across the entire production chain for maintenance purposes. The objective will be achieved by defining an integration methodology for heterogeneous models within a neural architecture. The approach will facilitate the exploitation of existing knowledge, the reuse of predictive models, and the reduction of the amount of data needed for model training. The approach will be demonstrated through the development of operational diagnostic systems on three industrial use cases.

  • Provision of services, Pilot Line and infrastructure
    The project will make use of the Pilot Line and the services offered by BI-REX: in particular, it is planned to rent the Line and use the Project Management service. This service is structured in this case through activities of: Coordination and Administrative Management (definition of working subgroups, scheduling conference calls, monitoring the state of work, management of deliverables); Communication and Dissemination (definition of the plan for the preparation and dissemination of material – news, videos, press releases, etc.. – to be conveyed on various channels – website, social networks, etc.). – plus organisation and promotion of scientific workshops/conferences).
  • Sharing, use, dissemination of product material and know-how
    The contents/materials produced may also be used by BI-REX in connection with its own training activities, with a view to dissemination and diffusion of know-how.
  • Bonfiglioli SpA (Coordinatore)
  • Alascom Srl
  • ENI SpA
  • Aetna Group SpA
  • MindIT Srl
  • Nier Ingegneria SpA
  • IMA SpA
  • Marposs SpA

Call: Call 1

Area: Ict for machines and production lines

Sub Area: Predictive Diagnostics based on Data Analytics and Machine Learning Techniques

Implementation period: 16.07.2020 – 16.01.2022