基于近红外光谱技术的玉米品种鉴别方法研究  

Research on identification methods of maize varieties based on near-infrared spectroscopy

作　　者：蔡月芹[1] 赵文杰 王雪[1] Cai Yueqin;Zhao Wenjie;Wang Xue(Heilongjiang Bayi Agricultural University, Daqing 163319, China)

机构地区：[1]黑龙江八一农垦大学

出　　处：《江苏科技信息》2019年第21期36-38,57,共4页Jiangsu Science and Technology Information

基　　金：黑龙江省教育厅科研项目;项目名称：基于光谱分析与神经网络的玉米品种识别技术的应用研究;项目编号：12541592

摘　　要：文章提出了一种利用近红外光谱技术与人工神经网络结合鉴别玉米品种的快速、无损的方法。收集3种常见东北种植的玉米品种共150个样本作为样本,在4000~10002cm-1范围内采集近红外漫反射光谱样本。经过卷积平滑法(Savitky-Golay)和去趋势校正预处理后,对数据进行主成分分析,再结合人工神经网络技术进行品种鉴别。模型对建模集120个样本鉴别率为100%,对预测集30个样本的鉴别率为100%。实验结果说明该方法能快速无损地鉴别玉米种子品种,为玉米种子的品种鉴别提供了一种新方法。In this paper, a rapid and nondestructive method for identifying maize varieties with near- infrared spectroscopy and artificial neural network is proposed.A total of 150 samples of three maize varieties from Northeast China are collected as experimental samples, and the near-infrared diffuse reflectance spectra of three maize seeds are collected in the range of 4 000~10 002 cm-1. After savitky-golay smoothing and detrend correction pretreatment, the data is analyzed by principal component, and the artificial neural network technology is used to identify the varieties. The identification rate of 120 samples in the model set is 100%, and the identification rate of 30 samples in the prediction set is 100%. The experimental results show that this method can identify maize seed varieties quickly and nondestructivly, and provides a new method for maize seed variety identification.

关 键 词：近红外光谱 人工神经网络 品种鉴别 

分 类 号：TN219[电子电信—物理电子学]