High tech production is often subject to small but significant variations in output quality that can affect yield, cost and revenues. These variations can be caused by known, specific factors (process inputs, personnel differences, etc.) or by unknown sources.
Automation systems may attempt to mitigate variation but often fail to do so due to lagging measurements, drifts in variation over time, and the complexity of its causes. Data can also be scarce, which adds to the challenge.
By adding SentianController to the automation system the system can make more accurate predictions of the process state at any given point to anticipate variation and optimize the production process accordingly. SentianController is very sample efficient and can deal with limited data. Moreover, SentianController’s variation reduction technology is compact, extremely fast, and can adapt to new data continually without being reset, making it possible to deploy in an edge solution. The result is an increase in yield and quality of up to 30%.