LIBS结合ANN对不同类型土壤中的Cu的定量检测

不同种类土壤的Cu浓度与其激光诱导击穿光谱(LI BS)强度之间存在不同的规律,本文利用LIBS结合人工神经网络(ANN)对土壤中的Cu进行 了定量分析,以实现不同种类土壤中Cu的定量检测。分别研究了单一类型土壤和3种土壤 类型基体下神经网络的应用情况。结果表明,在单一土壤基体情况下,应用反向传播(BP)神 经网络可以对土壤中 的Cu进行准确的检测,检测误差最大为10.17%；而在3种土壤基体下 ,BP神经网络的预 测准确度降低,检测误差不大于16%,并且线性神经网络对Cu浓度较 高的土壤样品预测准 确度较高,两种方法的检测准确度均高于内标法,BP神经网络能够更准确的描述单一土壤 类型的基体效应。LIBS结合ANN能有效解决土壤间存在的基体效应,LIBS结合ANN能有效解决 土壤间存在的基体效应,实现不同类型土壤中Cu元素的定量检测。

Quantitative detection of Cu in different types of soils using laser induced breakdown spectroscopy combined with artificial neural network

Different kinds of soils have differen t matrix effects,which leads to the fact that laws between the copper concentrations and their laser induced plasma spectroscopy (LIBS) are different. This paper quantitatively analyzes the copper concentrations in soils using LIBS combined with artificial neutral network (ANN),in order to realize the quantitative test of copper in different kinds of soils.The appl ications of the neutral network for one single kind of soil and three kinds of soils are studied respectively.In the case of a single soil matrix,the application of back propagation (BP) neutral network could measure the copper in the soils accurately,and the maximum measurement error is 10.17%.And for three kinds of soils,the prediction accurac y of BP neutral network is reduced,and the maximum measurement error is not larger than 16%.In this si tuation,the linear neural network has a higher prediction accuracy when the copper concentration is high ,and the two methods above both have better detection accuracy than the int ernal standard method.The laser induc ed breakdown spectroscopy combined with artificial neural network can effectively solve the matrix effect between different types of soils, achieve quantitative detection of copper,and may play a more important role in the field of soil heavy metal detection.