摘要
目的将L1范数约束的主成分分析(L1-PCA)法应用于近红外光谱对中药制剂的检测技术中,为近红外光谱快检技术提供新算法。方法以近红外光谱技术无损检测安胎丸为例,采用PCA法和L1-PCA法对其进行聚类,并对两种算法进行比较。结果运用两种算法,分别从安胎丸的近红外光谱图和指标成分的含量值两方面进行聚类,结果显示L1-PCA法比PCA法的聚类效果更佳。结论 L1-PCA法可以抑制光谱中非高斯噪声导致的某些波段的异常值,提高聚类的准确程度,为后期近红外光谱检测技术提供新算法。
Objective To use L1 norm-constrained principal component analysis(PCA) method in the detection of traditional Chinese medicine preparations by near infrared spectroscopy(NIRS), and to provide a new algorithm for NIRS rapid detection. Methods NIRS detection was used for Antai pills. Traditional PCA method and L1 norm-constrained PCA method were used to cluster pills of different years, and the two algorithms were compared. Results The two algorithms were used to cluster the NIRS spectra and the content values of the index components in the pills. The norm-constrained PCA method L1 was better than the PCA method. Conclusion The L1 normconstrained PCA method can suppress the outliers of certain bands caused by non-Gaussian noise in the spectrum, which also improves the accuracy of clustering, and provides a new algorithm for NIRS spectroscopy.
作者
马晋芳
葛发欢
肖环贤
郭拓
MA Jin-fang;GE Fa-huan;XIAO Huan-xian;GUO Tuo(School of Electrical Informatica and Artificial Intelligence,Shaanxi University of Science and Technology,Xi'an 710021;Guangzhou Pumin Information Technology Co. Ltd,Guangzhou 510006;School of Pharmaceutical Sciences,Sun Yat-Sen University,Guangzhou 510006;Jiangxi Poly Pharmaceutical Co. Ltd,Ganzhou Jiangxi 341900)
出处
《中南药学》
CAS
2019年第9期1451-1454,共4页
Central South Pharmacy
