基于太赫兹光谱的土壤中微塑料定量研究

Quantitative study of microplastics in soil based on terahertz spectroscopy

土壤中的微塑料对环境安全和人类健康都会产生负面影响。本研究以聚乙烯(PE)或聚氯乙烯(PVC)与土壤的混合物为研究对象,对样本在0.8~1.6 THz光谱吸光度数据分别进行归一化,均值中心化和移动窗口平滑预处理,以最小二乘支持向量机模型(LS-SVM)、极限学习机模型(ELM)和偏最小二乘模型(PLS)对土壤中微塑料进行预测。以决定系数(R2)和均方根误差(RMSE)作为模型性能的评价指标。由建模结果可知:移动窗口平滑处理后的LS-SVM模型精度最高,其R2大于0.9600,RMSE小于0.0060。数据经移动窗口平滑处理后PLS模型预测效果也值得认可,其R2大于0.9400,RMSE小于0.0080。本研究为监测农业土壤微塑料污染程度提供了新方法。

Microplastics have a negative impact on soil function,soil biodiversity,and human health.In this study,the mixture of polyethylene(PE)、polyvinyl chloride(PVC)and soil was used as the research object.The sample absorbance data at 0.8-1.6 THz were normalized,mean centered,and moving window smoothing preprocessing.The least square support vector machine model(LS-SVM),extreme learning model(ELM)and partial least square model(PLS)are used to predict the microplastics in the soil.The determination coefficient(R2)and root mean square error(RMSE)were used as evaluation indicators of model performance.According to the modeling results,it can be concluded that LS-SVM modeling after moving window smoothing was the best method for predicting the degree of microplastic pollution in the soil,producing an R2 value greater than 0.9600 and an RMSE less than 0.0060.After the data is smoothed by the moving window,the prediction effect of the PLS model is also worthy of recognition.Its R2 is greater than 0.9400,and the RMSE is less than 0.0080.This research provides a new way for predicting of the concentration of microplastics in the soil.