人工智能识别在金相检验中的发展现状与展望  

Development status and prospect of artificial intelligence recognition in metallographic inspection

作　　者：沈园竣 马飞跃[1,2] 谭廖嶙 SHEN Yuanjun;MA Feiyue;TAN Liaolin(Department of Utilities,CISDI Engineering Co.,Ltd.,Chongqing 401122,China;Department of Utilities,CISDI Thermal&Environmental Co.,Ltd.,Chongqing 401122,China)

机构地区：[1]中冶赛迪工程技术股份有限公司公用设计部,重庆401122 [2]重庆赛迪热工环保工程技术有限公司公用设计部,重庆401122

出　　处：《物理测试》2024年第3期59-63,共5页Physics Examination and Testing

摘　　要：金相检验是成品钢材质量控制的重要手段,通过对非金属夹杂物、晶粒度、显微组织、脱碳层、渗氮层等进行分类评级来反映质量问题,进而采取相应的措施优化冶炼、轧制生产工艺。金相检验过程完全依靠人工参照标准评级图谱进行,存在人为主观因素影响大、准确率低、效率低等缺点。人工智能技术的迅速发展,为金相自动识别评级提供了现实基础。概述了人工智能图像识别技术在金相检验中的应用现状和研究进展,讨论了其发展方向和面临的挑战。Metallographic inspection is an important approach for quality control of steel products.The quality problems can be reflected through classifying and rating of nonmetallic inclusions,grain size,microstructure,decarburized layer and nitrided layer.Then the corresponding measures are made to optimize the smelting and rolling production process.The rating process of metallographic inspection relies entirely on manual reference to the standard rating map,which has some disadvantages such as large influence of human subjective factors,low accuracy and low efficiency.The rapid development of artificial intelligence technology provides a realistic basis for the automatic identification and rating in metallography.The application status and research progress of artificial intelligence image recognition technology in metallographic inspection were summarized.Its development direction and challenges were discussed.

关 键 词：金相检验 人工智能 图像识别 自动评级 

分 类 号：TG142.1[一般工业技术—材料科学与工程] TP18[金属学及工艺—金属材料] TP391.41[金属学及工艺—金属学]