【摘要】 土壤沉积物中的重金属污染问题日益凸显，开发现场快速检测技术已成为确保污染监测有效性及推进环境治理不可或缺的手段。基于此，该研究提出了一种基于激光诱导击穿光谱技术（LIBS）结合机器学习算法的土壤沉积物重金属元素定量分析方法。首先，基于搭建的LIBS装置采集了土壤沉积物样本的光谱，探究了不同光谱预处理方法对光谱数据的预处理性能。紧接着基于变量重要性测量（VIM）对预处理后的光谱数据进行特征变量筛选。借助交叉验证对预处理方法、变量重要性阈值等参数进行了优化。基于优化的输入变量建立了土壤沉积物样本中3种重金属元素（Pb、Cu和Zn）的定量分析模型，并与其他校正模型的性能进行了比对。结果表明，该研究提出的VIM-RF校正模型表现出最佳的预测性能，对于Pb其R2p为0.993 0,RMSEP为0.029 8mg/kg，对于Cu其R2p为0.981 0, RMSEP为0.112 7mg/kg，对于Zn其R2p为0.992 0, RMSEP为0.166 2 mg/kg。由此可见，该文建立的方法有望为土壤沉积物环境重金属污染快速筛查及治理提供一定的理论参考依据。更多还原

【Abstract】 The issue of heavy metal contamination in soil sediment is becoming increasingly prevalent. The development of on-site rapid detection methods for heavy metal elements represents the only viable approach to achieving effective pollution monitoring and environmental governance. Accordingly, this study proposed a quantitative analysis method for heavy metal elements in soil sediments based on laser-induced breakdown spectroscopy combined with machine learning algorithms. Firstly,the spectral collection of soil sediment samples was completed using the constructed LIBS device,and the efficacy of various spectral preprocessing techniques on spectral data preprocessing was investigated. Subsequently,feature variable selection was conducted on the preprocessed spectral data,based on the measurement of variable importance. The preprocessing method,variable importance threshold,and other parameters were optimized using cross-validation. A quantitative analysis model for three heavy metal elements(Pb,Cu and Zn) in soil sediment samples was constructed based on optimized input variables. To further validate the performance of the model,a comparison was conducted with the performance of other calibration models. The results indicate that the VIM-RF calibration model proposed in this study exhibits the best predictive performance,with a R2p of 0. 993 0and a RMSEp of 0. 029 8 mg/kg for Pb,a R2p of 0. 981 0 and a RMSEp of 0. 112 7 mg/kg for Cu,and a R2p of 0. 992 0 and a RMSEp of 0. 166 2 mg/kg for Zn. It can be seen that the method established by this research institute is expected to provide a theoretical reference for the rapid screening and treatment of heavy metal pollution in soil sediment environments.更多还原