GOALS AND OBJECTIVES
To improve overall workshop efficiency.
To develop and implement a new machine learning-based (ML-based) recommendation service.
- A mathematical ML-model.
Previously, the composition and quantity of ferroalloys required for a particular grade of steel were determined manually, based on instructions and steelmakers’ own personal experience. This state of affairs led to a situation in which the consumption of ferroalloys was not always optimal.
Jet Infosystems specialists developed and implemented a service which is able to recommend optimal ferroalloy consumption during production. This solution was implemented at Novolipetsk Metallurgical’s first converter workshop.
The system uses a great amount of data to generate recommendations. This data includes: the target chemical composition for steel and ferroalloys, their interchangeability, the melting route, results from intermediate chemical analyses, the presence/absence of certain ferroalloys, as well as their cost and other historical data.
As present, this service has been implemented for the first converter workshop using an ordinary brand gauge.