基于太赫兹光谱的土壤重金属铅含量检测初步研究

发布时间：2018-02-22 21:46

本文关键词： 太赫兹时域光谱土壤重金属检测铅偏最小二乘法遗传算法　出处：《农业机械学报》2016年S1期 　论文类型：期刊论文

【摘要】：探索应用太赫兹时域光谱用于农田土壤重金属含量检测的机理和可行性,尝试发展一种新的检测方法和手段。研究了样品制备测量方法,确定了最佳的制备参数为:聚乙烯样品盒尺寸为35 mm×3.5 mm×35 mm,中间空隙1.5 mm;压片法样品质量220 mg,2.5 t压力;分别制作一定浓度梯度(30~900 mg/kg)的含铅土壤样品30组,使用Z-2型太赫兹时域光谱仪采集相应的土壤样品太赫兹透射光谱数据,对光谱数据进行预处理,包括平滑处理、多元散射校正、基线校正,采用全谱-偏最小二乘法、区间-偏最小二乘法和遗传算法-偏最小二乘法建立土壤样品中重金属元素铅含量的定标模型并对预测集样本进行预测。实验结果表明,采用样品盒法制作土壤样品并且基于遗传算法进行特征波段选择、结合偏最小二乘法的建模效果最佳,含重金属铅的土壤样本模型的标定集和预测集的相关系数分别为0.86和0.81,标定均方根误差和预测均方根误差分别为23.55 mg/kg和39.52 mg/kg。本研究通过分析土壤样品的太赫兹透射光谱吸收系数与重金属元素铅含量的相关关系,建立了较好的预测模型。
[Abstract]:To explore the mechanism and feasibility of using terahertz time-domain spectroscopy (THz) to detect heavy metal content in farmland soil, and to develop a new method and method for the determination of heavy metals in farmland soil. The optimum preparation parameters were determined as follows: the size of polyethylene sample box was 35 mm 脳 3.5 mm 脳 35 mm, the middle void was 1.5 mm; the pressure of the sample was 220 mg / m ~ (2.5 t) by pressing method; and the 30 groups of leaded soil samples with a certain concentration gradient of 30 ~ 900 mg 路kg ~ (-1) were made respectively. The THz transmission spectrum data of soil samples were collected by Z-2 terahertz time-domain spectrometer. The spectral data were preprocessed, including smoothing, multivariate scattering correction, baseline correction, full spectrum partial least square method. Interval-partial least square method and genetic algorithm-partial least squares method are used to establish calibration models of heavy metal lead content in soil samples and to predict the samples of heavy metal elements in soil samples. The method of sample box is used to make soil sample and the characteristic band is selected based on genetic algorithm, and the modeling effect of partial least square method is the best. The correlation coefficients of calibration set and prediction set of soil sample model containing heavy metal lead were 0.86 and 0.81, respectively, and the calibration root mean square error and predicted root mean square error were 23.55 mg/kg and 39.52 mg / kg 路kg respectively. In this study, terahertz permeability of soil samples was analyzed. The correlation between the absorption coefficient of emission spectrum and the lead content of heavy metal elements, A better prediction model is established.
【作者单位】：北京农业智能装备技术研究中心;农业部农业信息技术重点实验室;北京农业信息技术研究中心;
【基金】：国家自然科学基金项目(31301237) 北京市自然科学基金面上项目(6132009) 北京市农林科学院重点实验室建设项目(KJCXPT2016-04);北京市农林科学院级创新团队项目(JNKYT201604)
【分类号】：X833

【相似文献】