分析测试学报2025，Vol.44Issue(6)：1161-1168,8.DOI:10.12452/j.fxcsxb.241027487
基于XRF的CARS-GAF-MobileNet铝合金牌号分类研究
Research on CARS-GAF-MobileNet Aluminum Alloy Grades Classification Based on XRF
摘要
Abstract
Aluminum alloys are widely used in industry due to their excellent characteristics,and accurate classification of aluminum alloys grades can further promote the development of manufactur-ing and other fields.In this paper,a new aluminum alloy X-ray fluorescence(XRF)spectral classifi-cation framework CARS-GAF-MobileNet(CGM)was proposed.First,an XRF spectrometer was used to obtain XRF spectral data of the aluminum alloy samples.Then,a multi-element calibration-based competitive adaptive reweighted sampling(CARS)was proposed to select variables for the da-ta.Next,the one-dimensional spectra were converted into two-dimensional spectral images using Gramian angular field(GAF),and the grayscale images were converted into RGB images by color mapping.Finally,the converted 2D spectral images were inputs to the Mobilenet-V3 model to classi-fy the aluminum alloy samples.The experimental results showed that the final classification accuracy of the proposed CGM framework could reach 94.3%,which could accurately identify aluminum alloy samples of different grades.The CGM is a promising framework for identifying aluminum alloy grades,and it has superior theoretical guidance and application value for the aluminum alloy classifi-cation problem.关键词
X射线荧光/铝合金分类/格拉姆角场/竞争性自适应重加权采样/深度学习Key words
X-ray fluorescence/aluminum alloy classification/Gramian angular field/competi-tive adaptive reweighted sampling/deep learning分类
化学引用本文复制引用
吕树彬,万优,李福生,杨婉琪..基于XRF的CARS-GAF-MobileNet铝合金牌号分类研究[J].分析测试学报,2025,44(6):1161-1168,8.基金项目
国家自然科学基金资助项目(62075028) （62075028）
