Wolfram Computation Meets Knowledge

Wolfram Archive

Machine Learning Optimizes Automated Assembly Lines

Published January 9, 2006
Optimize automated assembly-line production processes.
Use machine learning framework and Mathematica to create adaptive models from data provided by the shop-floor monitoring system.
Automatically create interpretable computational models
Maximize overall equipment efficiency and product quality
Deepen insights into complex automation processes for continual improvement
The Mathematica Edge
Declarative programming language simply describes machine learning tasks
Combines tools for machine learning with statistics and mathematical modeling
Seamless integration with automation software platforms

Increasingly, our most essential products–such as cars, electronics, and home and work furnishings–are made by automated processes. It is impossible to reset these complex systems without the right decision support or automated recovery. Determining this critical information calls for machine learning.

The Mathematica application machine learning framework (MLF) by developer uni software plus is an innovative solution for these systems. MLF enables machines to improve their own processes based on analysis of past event data and other statistics, and helps to create models that are both understandable and computationally fast-paced.

MLF is an integral part of production systems for major manufacturers who rely on its data mining and modeling capabilities. Companies such as AMS Engineering–a system provider for highly automated assembly lines that counts Bosch, Braun, and Moeller among its dedicated customers–use MLF to improve overall equipment efficiency and manufacturing processes.

A given assembly line can easily involve more than 30 processing modules with hundreds of parameters, which change with each frequent product redesign. Mathematica‘s comprehensive descriptors and solvers combine with MLF‘s fast model creators and evaluators, accounting for factors such as the product design, equipment availability, production efficiency, and quality rate to continually improve machine “intelligence.”

Mathematica and MLF are used throughout the automated assembly process, from creating and testing the right models offline to being an integral part of the shop-floor management systems during production. “The power of Mathematica as a comprehensive platform is still underestimated,” says Herbert Exner, president of uni software plus. “The hybrid system lets us easily program complex tasks, solve for results, and seamlessly link to other environments. This is how we have designed machine learning framework.”