【摘要】 利用近红外光谱技术实现了大米中含水量的快速测定。为进一步提高近红外模型的精度和稳定性,采用4种波长筛选方法:遗传算法(GA)、无信息变量消除法(UVE)、无信息变量消除-遗传算法组合法(UVE-GA)以及连续投影算法(SPA),对大米水分近红外光谱特征波长进行了优选,并基于筛选出的光谱变量建立了大米含水量偏最小二乘法(PLS)模型。结果表明,相较于全光谱建模,4种特征波长优选方法不仅提升了所建模型的预测性能和精度,还有效地减少了建模时的光谱信息量,节省了建模时间;其中经过UVE-GA算法从全波段1154个波长中筛选出的68个特征波长建立的模型效果最好,其预测相关系数和预测均方根误差分别为0.9675和0.3915。综上所述,近红外光谱技术结合UVE-GA光谱处理方法能够实现大米水分含量的快速无损检测,为大米含水量的监督检测提供了技术依据。更多还原

【Abstract】 The moisture content of rice was analyzed by near infrared(NIR) spectroscopy rapidly. In order to optimize the accuracy and stability of the near-infrared model, four different wavelength variable selection methods, including GA, UVE, UVE-GA and SPA were applied to select effective wavelength variables of the NIR spectroscopy. Results show that, the four methods can effectively reduce the number of variables and improve the performance of the model. Among them, iPLS model using UVE-GA on 68 characteristic variables selected from full-spectrum which had 1154 wavelengths achieved the optimal performance. The correlation coefficient and RMSEP of the model were 0.9675 and 0.3915, respectively. In a word, the optimal near-infrared model established in this paper can be used for the analysis of content in rice samples rapidly and eco-friendly.更多还原