Create new values for the Swedish metal industry through the use and development of recent data analysis techniques for finding currently hidden relationships between known and unknown process parameters and the resulting product quality.
About the project
In other sectors, where such advanced computer-based data analysis is already in use, for example internet-based retail and services, and financial analysis, business have gone through disruptive changes, allowing for a next level of competitiveness. In areas with very complex tasks, like text and speech understanding, machine-translation, image understanding, and autonomous vehicles, there have been dramatic advancements. However, such advancements have not yet been seen in manufacturing. In metal manufacturing, specifically, a range of complex physical and chemical processes take place, processes that have yet to realize the potential of advanced data analysis. With this project, we expect to better understand the complexity of production processes, for better predictability and controllability of output quality.
In this project, we apply and advance machine learning techniques, for complex industrial manufacturing cases provided by Sandvik Materials Technology and SSAB. For these industrial cases, the process output quality is not fully predictable, but depends on a number of factors that are unknown today. The expectation is to control complexity and to find many opportunities for both quality, efficiency and sustainability improvements. Further, we expect to develop novel knowledge in improved algorithms for process complexity analysis and complexity management, thereby advancing the field of data analysis in general and machine learning in particular.
The application is for a three-year project, with a total budget of 5.77 MSEK, where 2.80 MSEK is applied for at the KK foundation, 2.80 MSEK is in-kind funding by Sandvik Materials Technology and SSAB, and 0.163 MSEK is co-funding from the University.