Predictive Quality (Ceramics Client)
Problem
- Many quality defects go undetected before tiles are furnaced
- Typically defects become visible only after the furnace at the visual quality inspection station
- Most of the defected products stay on the line and are processed unnecessarily after the defect occurs
- Due to sublte type of defects, overall production is impacted from waste of material, energy and workforce
Action
- Predict quality defects through anomaly detection using IoT data of certain production stages (e.g., surface temperature of ceramic tiles)
- Detect quality defect before the visual quality inspection stage, preventing costly rework and waste.
Tool Stack
- GCP (Big Query, Cloud Functions, Cloud Scheduler, Data Studio), Python
15
%Reduced Quality Defects
15
%Overall Equipment Efficiency Increase
5
MinsPrediction Resolution