【摘要】 应用近红外透射光谱技术,并结合化学计量学方法,对3种不同品牌的藿香正气水的质量进行鉴别。分别采用小波变换和遗传算法对原始近红外光谱进行数据压缩和波长选择,处理后的光谱数据用主成分分析法进行处理,所有17个藿香正气水样品按生产厂家得到正确分类。将光谱数据用不同方法前预处理,进而利用反传人工神级网络的学习预测能力建立识别模型。交叉验证结果表明,原始光谱、平滑光谱、小波变换及遗传算法预处理后建立的校正模型分类样品的百分比均达100%。小波变换预处理的建模速度最快,预测结果最好。更多还原

【Abstract】 Near infrared(NIR) transmission spectroscopy was applied to the discrimination of three brands of Huoxiang Zhengqi Tincture with the aid of chemometrics.Near infrared spectra data was compressed by wavelet transform(WT) and the optimal wavelengths were selected by genetic algorithm(GA).Then the principal component analysis(PCA) was chosen to analyze the reduced data to obtain well-grouped results.The NIR data was further predicted with different supervised pattern recognize method,back propagation-artificial neural network(BP-ANN),and recognition rates of 100% were achieved.It indicated that the model set up by the combination of WT technology and BP neural network could be the rapidest in analysis and precise in discrimination of Huoxing Zhengqi Tincture.更多还原