Abstract
The recent focus on sustainability and improved efficiency requires innovative approaches in industrial automation. We present SemKoRe, a knowledge graph developed to improve machine maintenance in the industrial domain. SemKoRe is vendor-agnostic, it helps Original Equipment Manufacturers (OEMs) to capture, share and exploit the failure knowledge generated by their customers machines located around the world. Based on our interactions with actual customers, it usually takes several hours to days to fix a machine-related issue. During this time, production stops and incurs cost in terms of lost production. SemKoRe significantly enhances the maintenance process by reducing the failure diagnostic time, and by centralizing machine maintenance knowledge fed by the experts and technicians around the world. We developed flexible architecture to cover our customers’ varying needs, along with failure and machine domain ontologies. To demonstrate the feasibility of SemKoRe, a proof-of-concept is developed. SemKoRe gathers all failure related data in the knowledge graph, and shares it among all connected customers in order to easily solve future failures of the same type. SemKoRe received the approval of several substantial clients located in USA, UK, France, Germany, Italy and China, associated with various segments such as pharmaceutical, automotive, HVAC and food and beverage.
Highlights
Industrial Internet of Things (IIoT) has emerged as an enabler of the rapid integration of advanced technologies in the industrial world [1]
Consider reusing existing ontologies: No existing machine domain or machine failure ontology was found for reuse, we developed both for this work
Even today most of the solutions on the market focus on the enterprise and IT side than on the operational side of large industries. This means that there are mature solutions that use semantic web technologies to bridge siloed enterprise data in RDBMS (Relational Database Management System) and unstructured data like documents but there is no mature solution that can do the same for data described in operation technology protocols, e.g., OPC-UA
Summary
Industrial Internet of Things (IIoT) has emerged as an enabler of the rapid integration of advanced technologies in the industrial world [1]. Failure details are usually captured manually, e.g., using spreadsheets (e.g., Excel) This approach is not fault proof as each person filling out the sheet cannot be expected to provide all the required information and even if that information is given, there will be semantic mismatch, e.g., one person describes issue as “abnormal rotation speed” while another person describes the same issue as “irregular spinning rate”. SemKoRe helps to avoid this semantic mismatch, and captures all the machine, failure and maintenance data as a knowledge graph, allowing several actors to benefit (Sections 4 and 5). To support the needs of different actors, SemKoRe is developed using the semantic web and ontologies [5] This approach helps to accommodate future requirements and to provide a clear separation of concerns between the application needs and the domain knowledge which in this case is machine domain and failure domain knowledge.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
