基于多阶导数拉曼光谱组合技术的矿物油模式分类

为了实现对法庭科学领域重质矿物油物证的快速、准确、无损的鉴定，该文基于光谱分析技术提出了一种多阶导数光谱数据组合分析的方法。收集了80种不同型号、不同厂家的重质矿物油样本，利用傅里叶变换拉曼光谱分析法采集样本的原始光谱数据和导数光谱数据，并通过结合化学计量学构建分类模型。在构建的主成分分析(PCA)结合径向基函数神经网络(RBF)分类模型中，对单独的原始光谱、一阶导数谱和二阶导数谱数据的训练集准确率分别为80.0%、86.7%和86.2%，测试集准确率分别为73.3%、80.0%和72.7%；对组合后的原始光谱+一阶导数谱、原始光谱+二阶导数谱和一阶导数谱+二阶导数谱数据的分类中，训练集准确率分别为97.0%、96.7%和100%，测试集准确率分别为85.7%、90.0%和100%。结果表明，对组合后的导数光谱与原始光谱构建分类模型，准确率更高。其中，基于一阶导数谱+二阶导数谱数据构建的PCA结合RBF分类模型的结果最为理想，准确率达100%。而K最近邻算法模型由于受到样本不均匀的影响，整体分类准确率均较低。利用组合的导数光谱与原始光谱数据构建分类模型能够实现对重质矿物油样本的快速、准确、无损鉴别，可为光谱组合技术在法庭科学及其他分析测试领域的应用提供一定的借鉴和参考。

Classification of Mineral Oil Patterns Based on Multi derivative Raman Spectral Fusion Technique

In order to realize the rapid,accurate and nondestructive identification of heavy mineral oil evidence in the field of forensic science,a multi-derivative spectral data combination analysis method based on spectral analysis technology was proposed in this paper.The spectral data for 80 kinds of heavy mineral oil samples of different models and manufacturers were collected by Fourier transform Raman spectral analysis method,and the classification models were constructed by combining the stoichiometry.In the constructed classification model of principal component analysis(PCA) combined with radial basis function neural network(RBF),the classifications of single original spectrum,first derivative spectrum and second derivative spectrum data were presented.The classification accuracies for the training set were 80.0%,86.7% and 86.2%,while those for the test set were 73.3%,80.0% and 72.7%,respectively.In the classifications of original spectrum-first derivative spectrum,original spectrum-second derivative spectrum and first derivative spectrum-second derivative spectrum after combination,the accuracies for the training set were 97.0%,96.7% and 100%,and those for the test set were 85.7%,90.0% and 100%,respectively.Results showed that the classification accuracy was higher when the derivative spectra and ordinary spectra were constructed.Among them,the classification model of PCA combined with RBF based on first derivative spectrum and second derivative spectrum data was the most ideal,with an accuracy of 100%.However,the model of K nearest neighbor algorithm had a low classification accuracy due to the influence of uneven samples.Construction of a classification model by the combination of derivative spectrum and the original spectral data could realize the fast,accurate and nondestructive identification of the heavy mineral oil samples,which could provide a certain reference for the application of spectral combination technology in forensic science and other analytical testing fields.