论文部分内容阅读
应用近红外光谱技术实现除草剂胁迫下油菜叶片中脯氨酸含量的检测。对248个经过除草剂丙酯草醚处理后的油菜叶片,经过烘干、磨碎后进行光谱扫描。经过Savitzky-Golay平滑、变量标准化(SNV)、二阶求导预处理后,应用偏最小二乘法(PLS)建立脯氨酸含量的预测模型,同时提取有效特征变量作为神经网络(BPNN)和最小二乘-支持向量机(LS-SVM)的输入值,并建立相应的模型。用186个样本建模,62个样本预测。结果表明,最小二乘-支持向量机能够获得最优的预测效果,预测的相关系数(r)、预测标准差(RMSEP)和偏差分别为0.995,0.041和0.000。说明应用近红外光谱技术结合最小二乘-支持向量机能够定量获得油菜叶片中脯氨酸的含量。
Application of Near Infrared Spectroscopy to Detect Proline Content in Rape Leaves under Herbicide Stress. 248 rapeseed leaves after herbicide treatment were dried, ground and subjected to spectral scanning. After Savitzky-Golay smoothing, variable normalization (SNV) and second-order derivative preprocessing, PLS was used to establish the prediction model of proline content, and the effective characteristic variables were extracted as the neural network (BPNN) and minimum The input of LS-SVM is established and the corresponding model is established. With 186 samples modeled, 62 samples were predicted. The results show that the least squares support vector machine can obtain the best prediction results. The correlation coefficient (r), prediction standard deviation (RMSEP) and deviation of prediction are 0.995, 0.041 and 0.000 respectively. It shows that the content of proline in rape leaves can be obtained quantitatively by near-infrared spectroscopy combined with least squares support vector machine.