Materials’microstructures are signatures of their alloying composition and processing history.Automated,quantitative analyses of microstructural constituents were lately accomplished through deep learning approaches....Materials’microstructures are signatures of their alloying composition and processing history.Automated,quantitative analyses of microstructural constituents were lately accomplished through deep learning approaches.However,their shortcomings are poor data efficiency and domain generalizability across data sets,inherently conflicting the expenses associated with annotating data through experts,and extensive materials diversity.To tackle both,we propose to apply a sub-class of transfer learning methods called unsupervised domain adaptation(UDA).UDA addresses the task of finding domain-invariant features when supplied with annotated source data and unannotated target data,such that performance on the latter is optimized.Exemplarily,this study is conducted on a lath-shaped bainite segmentation task in complex phase steel micrographs.Domains to bridge are selected to be different metallographic specimen preparations and distinct imaging modalities.We show that a state-of-the-art UDA approach substantially fosters the transfer between the investigated domains,underlining this technique’s potential to cope with materials variance.展开更多
In the current study,an as-cast 26%Cr high chromium cast iron(HCCI)alloy was subjected to dry-sliding linear wear tests,under different loads.The loads were selected based on analytically computing the critical load(P...In the current study,an as-cast 26%Cr high chromium cast iron(HCCI)alloy was subjected to dry-sliding linear wear tests,under different loads.The loads were selected based on analytically computing the critical load(PC)i.e.,the load necessary to induce plastic deformation.The PC was calculated to be 15 N and accordingly,a sub-critical load(5 N)and an over-critical load(20 N)were chosen.The influence of increasing the load during the wear test was investigated in terms of the matrix microstructural behaviour and its ability to support the surrounding carbides.The morphological aspects of the wear tracks,and the deformed matrix microstructure adjacent and underneath the track was analysed by confocal laser scanning microscope(CLSM)and scanning electron microscope(SEM),respectively.No evidence of plastic deformation of the matrix was observed below PC.On the contrary,at loads equal to and higher than PC,the austenitic matrix plastically deformed as evidenced by the presence of slip bands.Electron backscattered diffraction(EBSD)measurements in terms of grain reference orientation deviation,and micro-Vickers hardness of the austenitic matrix indicated a deformation depth of about 40μm at the maximum applied load of 20 N.The active wear mechanisms during sliding were a combination of both adhesive and abrasive wear,although increasing the load shifted the dominant mechanism towards abrasion.This was primarily attributable to the increased propensity for carbide cracking and fracturing,combined with the inability of the hardened austenitic matrix surface and sub-surface to adequately support the broken carbide fragments.Moreover,the shift in the dominant wear mechanism was also reflected in the wear volume and subsequently,the wear rate.展开更多
文摘Materials’microstructures are signatures of their alloying composition and processing history.Automated,quantitative analyses of microstructural constituents were lately accomplished through deep learning approaches.However,their shortcomings are poor data efficiency and domain generalizability across data sets,inherently conflicting the expenses associated with annotating data through experts,and extensive materials diversity.To tackle both,we propose to apply a sub-class of transfer learning methods called unsupervised domain adaptation(UDA).UDA addresses the task of finding domain-invariant features when supplied with annotated source data and unannotated target data,such that performance on the latter is optimized.Exemplarily,this study is conducted on a lath-shaped bainite segmentation task in complex phase steel micrographs.Domains to bridge are selected to be different metallographic specimen preparations and distinct imaging modalities.We show that a state-of-the-art UDA approach substantially fosters the transfer between the investigated domains,underlining this technique’s potential to cope with materials variance.
基金The present work is supported by funding from the Deutsche Forschungsgemeinschaft(DFG,project:GU 2102/2-1).
文摘In the current study,an as-cast 26%Cr high chromium cast iron(HCCI)alloy was subjected to dry-sliding linear wear tests,under different loads.The loads were selected based on analytically computing the critical load(PC)i.e.,the load necessary to induce plastic deformation.The PC was calculated to be 15 N and accordingly,a sub-critical load(5 N)and an over-critical load(20 N)were chosen.The influence of increasing the load during the wear test was investigated in terms of the matrix microstructural behaviour and its ability to support the surrounding carbides.The morphological aspects of the wear tracks,and the deformed matrix microstructure adjacent and underneath the track was analysed by confocal laser scanning microscope(CLSM)and scanning electron microscope(SEM),respectively.No evidence of plastic deformation of the matrix was observed below PC.On the contrary,at loads equal to and higher than PC,the austenitic matrix plastically deformed as evidenced by the presence of slip bands.Electron backscattered diffraction(EBSD)measurements in terms of grain reference orientation deviation,and micro-Vickers hardness of the austenitic matrix indicated a deformation depth of about 40μm at the maximum applied load of 20 N.The active wear mechanisms during sliding were a combination of both adhesive and abrasive wear,although increasing the load shifted the dominant mechanism towards abrasion.This was primarily attributable to the increased propensity for carbide cracking and fracturing,combined with the inability of the hardened austenitic matrix surface and sub-surface to adequately support the broken carbide fragments.Moreover,the shift in the dominant wear mechanism was also reflected in the wear volume and subsequently,the wear rate.
