Predict Commodity Quality in Production
Ensuring your production process generates the highest quality product possible
Challenge
One of the world’s leading steel producers
Steel quality is assessed through a time-intensive examination in a laboratory, resulting in the production process adjustment involving expensive chemical compounds
Wanted the steel quality to be continuously predicted during the production process to cut costs and time
Developed and trained Machine Learning model to continuously predict steel quality
Overcame key challenge of matching data, related to a single production batch, to the dataset with properties of all input materials and production parameters and to the time structure of the production process
Solution
Value
Saved cost and time by doing away with laboratory-based quality assurance process
Model provides continuous and concrete guidance directly to the production staff on recommended measures (e.g. adjustment of production parameters, such as airflow) to ensure high quality of steel production without use of chemicals
Roles
Data Scientist, Data Engineer
Tools
Sector
Manufacturing