Bridging the Gap Between Domain Ontologies for Predictive Maintenance with Machine Learning

Predictive maintenance often relies on the continuous monitorization of equipment behavior, generally provided by sensors or by the very equipment. Additional data from management software, including which materials are being used and what processes are executed on the equipment can be used to enrich the data streams and ontologies can be used to bridge the gap between these different domains, while also facilitating the comprehension of the results obtained by the analytic methods applied to the data. Existing ontologies model these problems independently, and a holistic view that takes in consideration the temporal requirements of predictive maintenance is not yet available. This paper analysis existing ontologies and proposes a number of extensions that bridge the gaps between them, while meeting the time-sensitive requirements of the problem.