A Semantic Middleware Architecture for PLM-ERP-MES Integration Leveraging Knowledge Graphs and AI/ML in Discrete Manufacturing

Information Technology (IT) Data Management systems helps us to track the evolution of a product during different phases of Design, Manufacturing, Sales and Distribution. Huge amount of Data is created during this process in the different systems used by various departments in the company. The systems can be classified as PLM for Design and Engineering, MES for Manufacturing and ERP for Sales and Distribution. These are separate systems having their own unique Databases which leads to Data incongruencies and System Integration issues. Different data structures are generated like Bill of Materials (BOM), Routings, Work Order, MRP I & II, Inventory Update, etc. during the life-cycle of a product that has to be shared across multiple systems. Presently APIs like REST, SOAP, etc. or Middlewares like MuleSoft, Boomi, etc. are used for exchanging the data between systems. This paper discusses the present functionalities of these APIs and Middleware, and their associated limitations. An end-to-end scenario of motorcycle development is considered and IDEF0 is used to map and consequently understand which type of Data structures are created in which system. Further, this study describes how a combination of Knowledge Graphs (KGs) and Artificial Intelligence/Machine Learning (AI/ML) can provide advanced capabilities for the integration of these IT systems.