基于正交小波包的异福片近红外光谱分析

Determination of Rifampicin and Isoniazide Tablets with NIRS Based on Orthogonal Wavelet Packet

目的利用小波包变换(WPT)提取异福片样品近红外漫反射光谱的特征信息,建立用近红外光谱(NIRS)快速测定异福片中有效成分的新方法。方法运用偏最小二乘法(PLS)建立特征信息与利福平和异烟肼含量之间的定量校正模型,对预测集的样品进行含量测定。结果校正模型在小波包分解尺度为6时预测精度最好,预测集利福平和异烟肼的相关系数(rp)分别从原始光谱的0.981 72和0.979 89提高到0.990 09和0.991 48,同时预测集均方根误差(RMSEP)分别从0.007 66和0.006 84减小到0.005 89和0.005 36。结论正交小波包多尺度分析对近红外光谱具有较强的去噪和压缩能力,从而使PLS模型更具有代表性和稳健性,同时也提高了建模效率和模型的预测精度。

OBJECTIVE To develop a novel algorithm of optimum orthogonal wavelet packet transformation （WPT） with multi- scale analysis to extract feature of near infrared diffuse reflectance spectroscopy （NIRS） for determination of rifampicin and isoniazide in Rifampicin and Isoniazide Tablets. METHODS The partial least squares （PLS） model with extraction feature was advanced and it was used to determine prediction sets of Rifampicin and Isoniazide Tablets. RESULTS The optimum scale for the model was 6, The regression coefficients （rp） of prediction sets are 0. 990 09 and 0. 991 48,they are much better than those in origin spectra that are 0. 981 72 and 0. 979 89. The root mean square errors of prediction （RMSEP） of them reduced from 0. 007 66 and 0. 006 84 to 0. 005 89 and 0. 005 36. CONCLUSION Wavelet packet transformation with multi-scale analysis is an effective method for compressing the spectra data and reducing the noises in NIRS, which makes PLS model more typically and moderately. The model＇s efficiency and precision are improved.