基于特征谱线比与时间分辨激光诱导击穿光谱的金属老化等级评估  

Evaluation of metal aging grade based on characteristic spectral line ratio and time-resolved laser-induced breakdown spectroscopy

作　　者：白凯杰 董美蓉[2] 尚子瀚 蔡俊斌 卢伟业 陆继东[2] BAI Kaijie;DONG Meirong;SHANG Zihan;CAI Junbin;LU Weiye;LU Jidong(Guangdong Institute of Special Equipment Inspection and Research Shunde Branch,Foshan 528300,China;College of Electric Power,South China University of Technology,Guangzhou 510640,China)

机构地区：[1]广东省特种设备检测研究院顺德检测院,广东佛山528300 [2]华南理工大学电力学院,广东广州510640

出　　处：《冶金分析》2023年第7期29-36,共8页Metallurgical Analysis

基　　金：广东省特种设备检测研究院科技项目(2021CY-2-02,2022CY-1-02);中央高校基本科研业务费专项资金资助(2022ZFJH04)。

摘　　要：激光诱导击穿光谱技术(LIBS)能实现耐热钢服役状态的快速原位监测,为运行生产安全提供保障。实验利用便携式LIBS检测设备,探究基于LIBS的耐热钢老化等级评估优化建模方案。以典型耐热钢T91为实验试样,首先探究了合金元素Cr、Mn与基底元素Fe的LIBS谱线特性,发现其特征谱线强度比与耐热钢老化等级之间存在线性相关关系;随后,比对了不同延迟时间下耐热钢特征光谱的演化规律,分析时间分辨光谱矩阵所含的光谱信息;最后,将谱线强度比与时间分辨光谱矩阵结合作为建模数据源,耦合递归特征消除(REF)、线性判别式分析(LDA)的特征筛选方法及支持向量机(SVM)建立老化等级评估模型。结果表明,改进后的模型性能要优于单一延迟时间下所建立的老化等级模型,预测准确度可达96.18%。所得结果为LIBS耐热钢老化等级评估提供了优化方法和科学依据。Laser-induced breakdown spectroscopy(LIBS) can realize the rapid in-situ monitoring of the state of heat-resistant steel,and provide security for operation and production.A portable LIBS device was used to explore the optimization scheme of heat-resistant steel aging grade evaluation model based on LIBS.The typical T91 heat-resistant steel was used as the experimental sample.Firstly,the LIBS spectral characteristics of alloying elements Cr,Mn and substrate element Fe were investigated.It was found out that there was a linear correlation between the intensity ratio of characteristic spectral lines and the aging grade of heat-resistant steel.Then the evolution of the characteristic spectra of heat-resistant steel under different delay times was compared,and the spectral information contained in the time-resolved spectral matrix was analyzed.Finally,the spectral line intensity ratio and time-resolved spectral matrix were used as modeling data sources,and the aging grade evaluation model was established by combining the feature selection method of recursive feature elimination(REF),linear discriminant analysis(LDA) and support vector machine(SVM).The results showed that the performance of the improved model is better than the model was established based on single delay time,which can get 96.18% prediction accuracy.The results provided an optimization method and scientific basis for the LIBS heat-resistant steel aging grade evaluation.

关 键 词：耐热钢 便携式激光诱导击穿光谱 老化等级 特征谱线比 时间分辨光谱 

分 类 号：O657.31[理学—分析化学] TF764.2[理学—化学]