紫外光谱结合人工神经网络识别不同品种红薯淀粉  

Identification of different varieties of sweet potato starch by UV spectroscopy combined with artificial neural network

作　　者：魏泉增[1] 靳景贺 李秉昌 王国营 王德国[1] 宋应彪 WEI Quan-zeng;JIN Jing-he;LI Bing-chang;WANG Guo-ying;WANG De-guo;SONG Ying-biao(Key Laboratory of Biomarker Based Rapid-detection Technology for Food Safety of Henan Province,Xuchang University,Xuchang 461000,Henan,China;Hebei Qinhuangdao Agricultural Technology Extension Station,Qinhuangdao 066000,Hebei,China;Yuzhou Sanhuaitang Vermicelli Factory,Yuzhou 461670,Henan,China;Henan Lilishu Food Co.,Ltd.,Yuzhou 461670,Henan,China)

机构地区：[1]许昌学院,河南省食品安全生物标识快检技术重点实验室,河南许昌461000 [2]河北省秦皇岛市农技推广总站,河北秦皇岛066000 [3]禹州市三槐堂粉条厂,河南禹州461670 [4]河南粒粒熟食品有限公司,河南禹州461670

出　　处：《粮食与油脂》2024年第10期152-158,共7页Cereals & Oils

基　　金：河南省高校重点科研项目(23A550016)。

摘　　要：为了识别不同品种红薯淀粉,建立紫外光谱的红薯淀粉品种识别模型。采用单因素试验优化提取条件,对不同品种红薯淀粉提取液进行紫外光谱扫描,将原始光谱数据预处理后进行主成分分析,比较不同数据处理方法区分品种的效果,并进行聚类分析,利用人工神经网络建立识别模型。结果表明:最佳提取溶剂为甲醇,最佳超声时间为20 min;紫外光谱图有相似的吸收峰,但吸光度存在差异;以小波降噪后一阶求导处理数据进行主成分分析的品种识别效果最好。建立的人工神经网络识别模型对13个品种预测的准确率为100%。因此,紫外光谱结合人工神经网络模型可作为红薯品种淀粉快速、准确识别的新方法。In order to identify different varieties of sweet potato starch,a model for identifying sweet potato starch varieties using UV spectroscopy was established.The single factor experiment was used to optimize the extraction conditions,and the UV spectroscopy scanning was carried out on different varieties of sweet potato starch extract.After the original spectral data was preprocessed,principal component analysis was carried out to compare the effect of different data processing methods to distinguish varieties,cluster analysis was carried out,and artificial neural network was used to establish the recognition model.The results showed that the optimal extraction solvent was methanol,and the optimal ultrasonic time was 20 min.The UV spectroscopy had similar absorption peaks,but there were differences in absorbance.The best performance of principal component analysis for variety identification was achieved by processing data with first-order differentiation after wavelet denoising.The established artificial neural network recognition model had prediction accuracy of 100%for 13 varieties.Therefore,the combination of ultraviolet spectroscopy combined with artificial neural network models could be used as a new method for rapid and accurate identification of sweet potato starch varieties.

关 键 词：红薯淀粉 紫外光谱 主成分分析 人工神经网络 

分 类 号：TS235.2[轻工技术与工程—粮食、油脂及植物蛋白工程]