基于近红外光谱的小麦品质分类研究

Study on Classification of Wheat Quality Based on Near Infrared Spectroscopy

为了快速、简便、准确地鉴别小麦品质的类别,本研究提出了应用近红外光谱分析技术结合BP神经网络的鉴别方法对小麦进行品质分类。研究过程中对小麦样品的光谱数据进行了详细分析,采用马氏距离剔除了光谱数据中异常数据,并通过主成分分析说明利用近红外光谱鉴别小麦品质分类的可行性。为了提高所建模型的性能,采用SPXY算法对小麦样品进行合理的划分。并选取了一阶微分加归一化的预处理方法来处理光谱数据,消除无关信息和噪声对小麦光谱数据的影响。运用偏最小二乘法压缩光谱数据,减少了数据量,节省建模时间。最后采用BP神经网络方法建立了小麦品质分类模型。实验结果显示:模型的鉴别效果较好,对强筋样品识别的准确率高达94.4%,弱筋样品识别的准确率高达100%。实现了快速、准确地对小麦品质强筋和弱筋两类的鉴别,对小麦生产、市场交易及食品加工有着非常重要的意义。

In order to quick, easy and accurately identify the classification of wheat quality, this paper put forward using identification method of near infrared spectral analysis technology combined with BP neural network for classification of wheat. Samples of wheat were carried out a detailed analysis of spectral data in the process of research. First of all, Mahalanob distance was applied on spectral data filtering, which could eliminate abnormal spectrum. And principal component analysis explained that near infrared spectroscopy identified the feasibility of the classification of wheat quality. In order to improve the performance of model, SPXY algorithm was used for reasonable division of samples of wheat. Then the data processing method of first derivative and SNV were used to deal with spectral data, which could eliminate irrelevant information and noise on wheat spectral data. The method of partial least squares was used to compress spectral data, which could reduce the amount of data and save modeling time. Finally, BP neural network as the modeling method was used to establish identification model of wheat quality. Experimental resuh showed that： the model identification effect was good, the recognition accuracy of strong gluten samples was as high as 94.4%, the identification accuracy of weak gluten samples was as high as 100%, which had realized quickly and accurately classification between strong gluten and weak gluten wheat, which had a very important significance on wheat production, market trading and food processing.