基于可见近红外光谱技术的车蜡品牌无损鉴别方法研究

发布时间：2018-01-20 06:20

本文关键词： 车蜡 Vis-NIR光谱线性判别方法最小二乘支持向量机连续投影算法　出处：《光谱学与光谱分析》2014年02期 　论文类型：期刊论文

【摘要】：探讨了可见-近红外光谱技术快速无损识别不同品牌车蜡的可行性。实验一共获得104样本,其中40个样本(建模集)用于建立模型,剩余64个样本(预测集)被用于独立验证建立好的模型。基于五种不同品牌车蜡的可见-近红外光谱分别建立了线性判别分析(linear Discriminant Analysis,LDA)和最小二乘支持向量机(least square-support vector machine,LS-SVM)模型。基于两个算法的全波段光谱模型的预测集正确率分别达到了84%和97%。进一步采用连续投影算法(successive projections algorithm,SPA)算法从751波段中选取了7个特征波段(351,365,401,441,605,926和980nm)。基于SPA选择的变量建立LS-SVM模型,准确率依然保持在97%。说明SPA选择的特征波段包含了对于车蜡品牌鉴别最重要的光谱信息,而大多数无用信息则被有效剔除。将SPA与LS-SVM算法的车蜡识别模型在保证正确率的基础上,还可以大大降低模型计算复杂程度,说明该模型能快速准确的从车蜡可见-近红外光谱中提取有效信息,并实现车蜡品牌的无损鉴别。
[Abstract]:The feasibility of rapid and nondestructive identification of different brands of car wax by visible near infrared spectroscopy (NIR) was discussed. 104 samples were obtained, of which 40 samples (modeling set) were used for modeling. The remaining 64 samples (prediction sets) were used to independently verify the established models. Based on the visible and near infrared spectra of five different brands of vehicle wax, linear discriminant analysis (LDA) was established. Linear Discriminant Analysis. LDAs and least squares support vector machines (LDAs) and least square-support vector machine (LDAs). LS-SVM model. The prediction set accuracy of the full-band spectral model based on two algorithms reached 84% and 97 respectively. Successive projections algorithm. The spa algorithm selects seven characteristic bands from the 751 band. 926 and 980nmg. based on the variables selected by SPA, the LS-SVM model was established. The accuracy remains at 97. It shows that the characteristic bands selected by SPA contain the most important spectral information for the identification of car wax brands. However, most of the useless information is eliminated effectively. The model based on SPA and LS-SVM algorithm can greatly reduce the complexity of model calculation on the basis of ensuring the correct rate. It shows that the model can extract effective information quickly and accurately from the visible-near infrared spectrum of car wax and realize the nondestructive identification of car wax brands.
【作者单位】：浙江经济职业技术学院;浙江大学生物系统工程与食品科学学院;
【基金】：国家自然科学基金项目(31072247)资助
【分类号】：TP391.41
【正文快照】： 引言近年来,随着汽车工业的蓬勃发展,汽车保有量急剧增加,爱车族都在提高车子的外在品质上下功夫,给自己的爱车穿上一件光鲜亮丽的“衣裳”。汽车打蜡能够有效地保护漆面,延长汽车壳体的使用寿命,并使车辆外表光亮美观。车蜡是一种由蜡、表面活性剂、溶剂等组成的化合物,易于

【参考文献】