论文部分内容阅读
[目的]探索快速测定完整黍稷籽粒蛋白含量的方法。[方法]采用近红外光谱分析技术建立数学模型并进行预测,比较原始透射光谱经导数处理结合不同回归算法对模型的影响。[结果]分别经一阶和二阶导数处理后利用偏小二乘法和改进的偏小二乘法,4 种方法的分析效果相近,最优的是一阶导数结合改进的偏最小二乘回归法,黍稷蛋白定标模型的定标相关系数(RSQ)为0.880 6,定标标准误差(SEC)为0.342 4,交互定标标准误差(SECV)为 0.375 1,外部预测标准误差(SEP)为 0.454。[结论]以完整黍稷籽粒为样品所建立的蛋白 NITS 模型,可以用于黍稷蛋白含量的快速检测。
[Objective] The research aimed to explore a method for rapid determination of grain protein content in whole millet. [Method] The mathematical model was established and predicted by near infrared spectroscopy. The influence of the original transmission spectrum on the model by derivative treatment and different regression algorithms was compared. [Results] The analysis results of the four methods were similar by using the partial least square method and the improved partial least squares method after the first and second derivatives were processed respectively. The optimal ones were the first derivative and the improved partial least squares regression (RSQ) was 0.880 6, the SEC was 0.342 4, the SECV was 0.375 1, the standard deviation of external prediction was (SEP) was 0.454. [Conclusion] The protein NITS model established from the whole grain of Zea mays could be used to detect the protein content of.