MISE Projects

Predictive Maintenance and the Analysis of Production Processes

Predictive Maintenance

One of the projects developed by Travi e Profilati di Pallazeno team together with the Engineering Department of the University of Brescia deals with the predictive maintenance of the plant in San Zeno Naviglio through machine learning technics.

The project focused on data gathered from the fumes filtering systems’ fan: an algorithm has been developed to predict the evolution of different variables, comparing the predicted results with data from sensors.

Comparing data and outcome, the team developed a specific method to foresee and detect system failures.

Production Process Analysis

TPP team from Duferco Travi e Profilati and Catholic University of Brescia, developed a project of a platform for forecasting the chemical characteristics of the casting. In particular, thanks to the large amount of data available on historical production and a complex calculation application based on the Euclidean metric, it is possible to anticipate casting recipes by comparing the desired characteristics with the data in the archive.
The modularity and outstanding usability of the web interface makes the system extremely flexible: you can select the set of significant elements for each case, giving a weight to the different inputs entered and focusing the research on the most relevant aspects from time to time. The user is provided with information on the casting, the production process and the details of the ferroalloys added in similar castings carried out in precedence. The research can be refined according to filters related to the type of steel and the time frame of production.
A further step of the project, currently under development, will allow the user to search ever wider and more accurate, while simplifying the way to access data: operators can use a single heat ID as input for research, instead of individual values, inserted one by one. The database will also be expanded and enriched with new data sets that will improve the quality of the output.