Before
Fierce competition in the smartphone industry requires a high quality and low cost product which is manufactured efficiently. Failures rates need to be below 0.03% to be profitable.
After
Switched from the regular cutting tool replacement process at every 200 products to a monitored based system that extended the life of a cutting tool by up to 3 times.
The How
- 4 challenges identified and addressed using Data Analytics and Machine Learning
- Develop a quality predictor using Deep learning
- Improve the manufacturing process using Sensor data mining
- Improve the precision cutting process using Data driven modelling
- Optimise the process parameter using Reinforcement learning
- Combine the inferences from these models to predict when cutting tools need to be replaced
- Alert maintenance teams in advance to replace the tools
