CUSTOMER STORY | GLASS MANUFACTURING
AGC classifies float glass defects in under 12 seconds with more than 97% accuracy
AGC faces two data challenges on its float glass lines: microscopic defects below 100 µm, and furnaces at 1,400°C where sensors degrade. TIMi powers both solutions.
Defect classification accuracy
>97%
Processing latency, in time for cutting
<12 s
Fuel consumption
-5%
OVERVIEW
The challenge.
Two distinct and complex data challenges.
Defects below 100 µm
High-end industries such as aerospace, automotive and space have zero-tolerance thresholds, while others, like construction, are more flexible.
Furnaces at 1,400°C
At this temperature, sensors degrade rapidly and produce noisy, unreliable data.
SOLUTION
What was done.
A real-time classification engine, and a model of the furnace.
Real-time defect routing
Laser-scanned defects are classified into 12 categories in under 12 seconds, just in time for cutting.
Right sheet, right customer
With more than 97% accuracy, sheets are routed to the right customer, which reduces waste.
Furnace thermal model
The thermal dynamics of the furnace are modeled despite sensor degradation and data noise at 1,400°C.
RESULTS
What changed.
Accurate classification, and lower fuel consumption.
Classification
More than 97% accuracy, in under 12 seconds.
Waste
Sheets routed to the right customer, with less waste.
Energy
A 5% drop in fuel consumption.
SEE FOR YOURSELF
Test it on your own data.
Run TIMi on your own data with a free trial, or talk to us about your case.
