ndustry guided harnessing of materials big-data using machine learning

Purpose and goal

The coating industry faces with the need of materials that have high toughness for high-precision, high-temperature metal machining or high corrosion resistance for cathode materials in modern electrolyte batteries and fuel cells. An artificial intelligence based method will be utilized to tailor-explore the big data of materials and accelerate the search of multifunctional coatings. The project directly involves two world-leading companies in the area of hard coatings, Sandvik Coromant and Seco Tools, which provide access to computationally inaccessible control knowledge.

Expected results and effects

In the project we have built a materials database of industrially relevant nitride, carbide, oxide materials. We have served a software to serach for materials in the database and analyze their elastic properties. Machine learning algorithms has been developed and tested for two simple descriptors. Besides these results, which are directly utilized in FunMat-II competence center from now, a novel research idea for machine learning hardness has been established and transferred into FunMat-II.

Planned approach and implementation

The results has been developed at Linköping University mostly by a PhD student. Seminars, meeting days and long term exchanges have served the success of the project. Sandvik personal has spent a week exchange at Linköping University, while F.T has spent three months in Moscow to broader his experience in machine learning techniques.