Resources for AI in Manufacturing

your journey - from data to algorithms to use-cases

The application of AI to manufacturing environments embraces many stakeholders and builds on a large set of indispensable resources. Overall, you should look at two key topics, Data and Algorithms.

Data

the fuel of AI in manufacturing

Data is the fuel of AI-in-manufacturing. It contains valuable information from machines and products and allows us to observe the production. However, without understanding the data, we cannot understand what is happening on the shop floor from purely looking at the data.

When acquiring data for use in AI applications, it is key to consider quality of the data, annotation, and storage.

Quality

Quality of data is essential to enable meaningful interpretation and use. Key criteria are

  • does the data describe the underlying physical condition precisely enough (signal to noise ratio of acquired data)
  • is it clearly described what the data describes (provision of meta data)
  • has the data been acquired frequently enough to represent relevant events (adequate time resolution)

Annotation

Data needs to be identifiable. Guiding questions are

  • where does the data come from
  • when was the data acquired
  • what unit have the values

Storage

Storage of data is essential for use in analysis and control. Key decisions to be taken

  • is the data backed up (backup and redundant storage)
  • is the recorded data write-protected (immutability)
  • does the data contain personal data (increased level of data protection, GDPR)

If you want to discuss your challenges on data, check out our service offerings.

Service Offerings

Algorithms

the workhorse of applications

Algorithms are the workhorses of AI applications. They are designed and composed by experts to support people or systems to take decisions. Algorithms use the data and process the data to technically understand what happens in the manufacturing system.

AI technologies

  • machine learning
  • deep learning
  • reinforcement learning
  • unsupervised / supervised learning

Things to consider in the selection of data and use of algorithms are

  • how much compute resource will the application need
  • does the application yield stable results
  • has the result of the ai-based computing ethical aspects
  • does the application process personal data

If you have an idea on using AI in your manufacturing environment, then check out our service offerings.

Service Offerings

Use Cases

the proof of feasibility

Much has been demonstrated on AI in logistics and on AI in text creation and recently on AI in arts. In manufacturing, the application of AI still is not common, but shows great potential through successful demonstrations. Key targets are

  • reduction of scrap
  • optimisation of production processes
  • increase of resilience within the production chain
  • enhancement of product quality

check out the use-cases section