High tech manufacturing variation reduction
Read more at
https://www.sentian.ai/
One of the customers manufacturing processes experienced small but significant process variation. By adding AI to the automation and control system the goal was to reduce this variation and improve sensor accuracy and overall yield. The result was an increase (20%) in the proportion of sensors in the highest quality tier.
OPPORTUNITIES & CHALLENGES:
- Highly automated high-tech manufacturing, using advanced machines and many robots.
- Limited data availability
- Variation drifted over time
- Potential sources of variation were complex, incompletely understood, and in some cases not known when settings were made
Lag time of several weeks from process completion to lab results, which required sophisticated imputation methods
The goal was to reduce the variation and increase the quality of the produced sensors.
Product: SentianController
Features:
- Continuous process optimization
- Automated solution
- Handling of limited data volume (sample efficiency)
The solution managed to solve a problem that had been hard to solve for many years. It is a good example of when AI can find patterns in data that was too hard for even experts to solve.
- Increase (20%) in the proportion of sensors in the highest quality tier
- Handles updates to process automatically
- Applied to all similar production lines
Manufacturing
Production
Better Quality, More Efficient, Saving Cost
Shaper
Optimization
Machine Learning, Reinforcement
Time series, Sensor Data