论文部分内容阅读
为实现不同种类土壤的快速分类鉴别,实验研究了基于激光诱导击穿光谱技术的土壤快速分类方法。由于不同类型的土壤在元素组成上会存在较大差异,所以利用激光诱导击穿光谱技术进行土壤分类具有可行性。不同土壤在相同实验条件下产生的等离子体温度会存在较大差异,可以作为分类的重要依据,所选择的7类土壤中,赤红壤的等离子体温度最高。选取土壤中6种常量元素Si,Fe,Al,Mg,Ca和Ti的光谱强度作为分类指标,利用主成分分析(principal component analysis,PCA)对7种土类的25个样品进行了分类,其中砖红壤和赤红壤分类出现了交叠,而不同高山草甸土样品之间元素差异较大,并没有实现较好的聚类。利用反向传播神经网络(back-propagation artificial neural network)结合土壤的LIBS光谱对土壤进行了分类,分类结果与PCA结果相近,赤红壤与砖红壤出现了识别错误。当用PCA分析获得三个主成分值作为BP神经网络的输入量时,获得了较好的分类结果,因为简化了输入量,降低了BP神经网络的误差,此时只有一个高山草甸土被识别成褐土,而高山草甸土的等离子体温度显著低于褐土,所以结合不同土壤类型的等离子体温度差异,能够实现不同土壤的分类识别。实验证明激光诱导击穿光谱技术可以应用于土壤分类,为土壤普查和合理利用提高了一种新的技术。
In order to realize rapid classification and identification of different kinds of soils, a rapid soil classification method based on laser-induced breakdown spectroscopy was experimentally studied. Because different types of soils may have large differences in elemental composition, it is feasible to classify soil using laser-induced breakdown spectroscopy. The plasma temperature produced by different soils under the same experimental conditions will be greatly different and can be used as an important basis for classification. Of the seven selected soils, the plasma temperature of the latosol is the highest. The spectral intensities of six kinds of elements Si, Fe, Al, Mg, Ca and Ti in soils were selected as the classification index, and the 25 soil samples from 25 soil samples were classified by principal component analysis (PCA) The classification of brick red soil and latosolic red soil overlapped, while the elemental differences between different alpine meadow soil samples did not achieve good clustering. The soil was classified by using back-propagation artificial neural network combined with LIBS spectra of soil. The classification result was similar to that of PCA, and the recognition error was found between latosolic red soil and latosol. When using PCA analysis to obtain the three principal component values as the input of BP neural network, better classification results are obtained because the input is simplified and the error of BP neural network is reduced. At this time, only one alpine meadow soil However, the temperature of plasma of alpine meadow soil is significantly lower than that of cinnamon soil, so the classification and identification of different soils can be realized by combining the difference of plasma temperature of different soil types. Experiments show that laser-induced breakdown spectroscopy can be applied to soil classification, which improves a new technique for soil survey and reasonable utilization.