摘要太子参作为我国传统中药材被人们广泛的使用，它的质量问题尤其受到人们关 注，相应的，作为质量信息的一种，太子参产地信息显得格外重要。近红外光谱检 测方法作为一种新兴的在线即时检测技术正在被运用到越来越多的领域中,本实验 中将探究使用近红外光谱检测太子参产地的可行性。83570

本实验中，将使用便携式的近红外设备获得太子参的光谱信息，对福建、安 徽、贵州三个产地的太子参数据预处理后进行偏最小二乘法进行建模，结果显示， 福建产太子参的预测正确率为 91。7 %；安徽产太子参的预测正确率为 90%；贵州太 子参的预测正确率为 93。2%。

同时为了提高预测结果的准确性，去除冗长信息，对福建、安徽、贵州三个产 地的太子参数据使用随机蛙和自适应重加权采样的方法进行波长优选。结果表明在 鉴定不同产地的太子参时，不同的方法效果不一样。

在自适应重加权采样的方法下，福建产太子参的预测正确率为 95。8%；安徽省 太子参的预测正确率为 100%；；贵州产太子参的预测正确率为 97。7%。

而在随机蛙方法下，福建太子参的预测正确率为 91。7%；安徽太子参的预测正 确率为 100%；；贵州太子参的预测正确率为 100%。

与预处理的结果相比较，使用波长优选后，太子参产地的预测正确率有了明显 的提高；而在波长优选方法之间的比较中，福建省的太子参较为适合使用 CARS 方 法进行波长优选，安徽太子参两种方法都可以，贵州太子参使用随机蛙方法效果较 好。

综上所述，使用近红外光谱进行太子参产地的检测是可行的。

毕业论文关键词:太子参；近红外光谱；matlab；偏最小二乘法；随机蛙；自适应重加权采样

Abstract Heterophylla As Chinese traditional medicine is widely used by people, so people pay aattention to it’s quality especially; appropriateiy, as one of quality information, the information of heterophylla origin is particularly important。 Near infrared spectroscopy as a new method of instant online detection  technology is being applied to more and more areas, this experiment will explore the feasibility of detecting heterophylla origin。by using near-infrared spectroscopy。

In this study, using portable NIR instruments to obtain spectral information heterophally, modeling after preprocessing heterophally data of Fujian, Anhui,and Guizhou by using plsda method, the results show, the prediction accuracy of heterophylla produced in Fujian is 91。7%; prediction accuracy rate of Anhui heterophylla is 90%; prediction accuracy of Guizhou heterophylla was 93。2%。

Meanwhile, in order to improve the accuracy of predictions, removing redundant information, using random frogs and plsda methods to prefer sampling wavelength of heterophylla data ofFujian, Anhui and Guizhou。 The results show that in the identification of different origin heterophylla, a different approach is not the same effect。

In the plsda method, prediction accuracy rate of heterophylla produced in Fujian is 95。8%; Anhui heterophylla’s prediction accuracy rate is 100%; Guizhou heterophylla’s prediction accuracy is 97。7%。

In The random frog method, Fujian heterophylla’s prediction accuracy rate is 91。7%; Anhui heterophylla’s prediction accuracy rate is 100%; Guizhou heterophylla’s prediction accuracy rate is 100%。

Compared with the results of the pre-treatment, after the use of the preferred wavelength, heterophylla origin’s prediction accuracy has been significantly improved; while in the more preferred method of wavelength between, Fujian heterophylla more suits of CARS method Anhui heterophylla suits both methods, Guizhou heterophylla’s results by using random frog method is better。