压力容器2025，Vol.42Issue(5)：12-21,10.DOI:10.3969/j.issn.1001-4837.2025.05.002
基于自适应多特征融合的12Cr1MoV钢球化等级智能判定研究
Research on intelligent determination of spheroidization grade for 12Cr1MoV steel based on adaptive multi-feature fusion
摘要
Abstract
To address the issue of human bias in determining the spheroidization grade of 12Cr1MoV steel,a machine learning algorithm based on adaptive multi-feature fusion is proposed,along with the development of spheroidization image rating software.By collecting 263 metallographic image samples of pressure equipment made of 12Cr1MoV steel after long-term service(including 147 pearlite spheroidization images and 116 bainite spheroidization images),a recognition model was constructed by integrating Canny edge features,fractal dimension features,and deep learning features.The results show that for pearlite spheroidization images,the recognition accuracy of the adaptive multi-feature fusion model reaches 93%,which is 39%higher than that of the traditional Canny-XGBoost method.For bainite spheroidization images,the accuracy is 90%,representing a relative improvement of 42%.This study significantly improves the accuracy and efficiency of spheroidization grade determination,achieving intelligent evaluation of spheroidization grades under small-sample conditions,which is of great significance for ensuring equipment safety.关键词
12Cr1MoV钢球化/自适应多特征融合/机器学习/智能评级Key words
12Cr1MoV steel spheroidization/adaptive multi-feature fusion/machine learning/intelligent rating分类
机械制造引用本文复制引用
王海斌,程经纬,董杰,王家辉..基于自适应多特征融合的12Cr1MoV钢球化等级智能判定研究[J].压力容器,2025,42(5):12-21,10.基金项目
合肥通用机械研究院有限公司青年基金项目(2022011502) （2022011502）
