General Electric is working to make AI an integral component of its industrial processes, from design (and co-design) to production (and co-production) to operation (and co-operation). Notice the (co-) part. It is an important aspect that moves GE into industry 4.0 and it is notable that they are looking at AI as a fundamental tool.
For each of their products they have a digital representation that through AI embeds the “semantics” of each parts (what they do, how they work together, what are their constraints…). This digital representation (they called them “digital twins”) can be used for simulation, for sharing with other companies for monitoring the data coming from operation. AI is used to “understand” what is going on, to detect anomalies and prevent issues.
An engine on board an aircraft flying from London to New York may have its digital twin residing in a cloud in Los Angeles. Notice that the concept of digital twin applies to each instance of a product, each engine has its own digital twin that has operated (although virtually) exactly the same hours and sustaining the same acceleration and mechanical stress of its sibling up in the air.
To this effect GE has started to train its employees in machine learning, 400 have already been certified and many more will follow.
The simulations run on the digital twins provide guidance on the “atom” twin. By applying AI to the digital twin and then sending instruction to the atom twin GE is able to increase productivity of wind farms by 20% and decrease fuel consumption of diesel engines in a locomotive by 32,000 gallon per year.
Notice here one of the typical characteristics of Industry 4.0. A continuous relation between production and operation.
Engineers at GE are learning AI to understand what machines can learn and how they can interact, but this is also reshaping their relation with the machines. They are moving towards a symbiotic relationship with machines at the various stages of design, production and operation.
Industry 4.0 with the pervasive presence of autonomous robots and distributed AI will be one of the first environment populated by symbiotic autonomous systems. Machines will learn to rely on humans and humans will learn to rely on machines, each one adapting dynamically along the way.