Universal Machine Learning System for Material Properties Prediction

The importance of accurate material properties information for engineering calculations and simulations, such as CAE (Computer Aided Engineering) and FEA (Finite Element Analysis), can never be overstated. For example for structural steels, conventional mechanical properties such as yield strength, tensile strength, hardness, and ductility may vary more than tenfold at room temperature, depending on the variations of the alloying elements, heat treatments, and fabrications. With even a moderate change in working temperatures, the properties’ variations and changes can become even more profound, and approximating using typical property values for some groups of alloys may lead to very serious errors.

While large material information sources can help with these challenges, sometimes there is missing information that is not covered by standards, producer data sheets, and other experimental sources. However, recent developments in artificial intelligence and machine learning provide an opportunity to overcome these gaps.

This paper presents a machine learning system aimed at predicting material properties of a wide range of diversified materials, such as stainless steels, aluminum, coppers, refractory alloys, and polymers. By using copious training sets provided by a materials database with over 25M properties and a proprietary methodology for taxonomy, data curation, and normalization, this developed system can predict physical and mechanical properties for thousands of materials at various temperatures, heat treatments, and delivery conditions. The accuracy achieved in the terms of relative error in most cases is above 90%, and frequently above 95%.