Lityum-iyon Bataryaların Katot Malzemelerindeki Kristal Yapının Kestirimi için Makine Öğrenimi Yöntemlerinin Uygulanması

Abstract

It has commonly been assumed that the physical and chemical characteristics of lithium-ion silicate cathodes are influenced significantly by the crystal structure system. Because of this, crystal structure estimation has played a vital role in bringing about forecasting many other features of cathodes in battery applications. Using a variety of classification techniques in machine learning which three primary crystal structure (monoclinic, orthorhombic, and triclinic) of silicate-based cathode materials with compound systems of Li-Si-(Mn,Fe,Co)-O has been estimated. The computations are based on the Materials Project's density functional theory computations. In this study, it has been explained that based on property evaluation in statistical models, the considerable correlation between the crystal system and other physical characteristics of the cathodes was validated. Furthermore, the parameters of several categorization techniques have been tuned in order to achieve maximum prediction accuracy. This case has shown that the XGBoost and Support Vector Classifier algorithms carried out in this study the highest forecasting accuracy in this study along with many other classification methods in cross-validation tests.