Enabling Machine Learning for Manufacturing Machines

The shop floor is full of machines and none of them are learning! Although machine learning languages and systems can handle complex manufacturing data, they aren’t widely utilized for automated manufacturing. A missing link is a framework for sharing, learning from mistakes, and optimizing results.

ISO 23247 describes a standard for Digital Twin manufacturing. In this framework, physical manufacturing elements (PME’s) are connected to data collecting and controlling elements (DCCE’s). The physical manufacturing elements are robots and machine tools. The DCCE’s are hosted on edge computing devices. Each DCCE controls and models the results of a manufacturing process. At the end of the manufacturing, as the plane rolls off the assembly line, its Digital Twin hops onto a USB.

This presentation will discuss Digital Twin manufacturing, including:

  • Managing the extremely large quantities of data of large products.
  • Dealing with complexity and subtle dependencies between properties
  • Data updates in real time to keep the digital twins current with the physical.

And the benefits, including:

  • The ability to rapidly adapt to engineering changes, to work with many vendors, to share results, and to grow from small to large.
  • Integration into systems used by the machine learning communities.
(CAMSC) Computer Aided Manufacturing & Supply Chain