摘要
快速检测生姜含水率对生姜的存储加工和国际贸易非常重要。本文应用可见近红外光谱快速检测生姜含水率，采集330个生姜的可见近红外光谱（光谱范围350-1800mm），然后用烘干法测定生姜的含水率，把330个生姜样本按照含水率的大小以2：1的比例分成校正组和预测组。应用专业知识法、偏最小二乘法和遗传算法j种光谱选择方法建立生姜含水率的预测模型，其模型的精度比应用全光谱（包含1451个光谱变量）所建立的模型精度高。通过比较，应用遗传算法所得预测模型的效果最好，选定的光谱数和因子数分别是300和6，预测组的相关系数、均方根误差和分别是0．9900和4．4440。
Accurate measurement of ginger moisture content （MC） is critical in marketing, storing, and processing. Dried ginger is likely the most acceptable form of ginger in the local and international market. In this study, Vis/NIR spectroscopy （range from 350 to 1800nm） was used to measure the moisture content of ginger. 330 samples were separated into two groups, as training and validation. Three different approaches for selection variables used to establish the PLS model were tested and compared： variable selection based on expert knowledge, interval PLS and genetic algorithm PLS. In comparison to the full spectrum model （contained 1451 variables）, the prediction capability was improved after using variable selection for PLS models and all three variable selection approaches gave similar results. By considering the minimum number of variables and latent variables （LVs）, of all the four PLS models, the application of the GA-PLS model could obtain the satisfactory result in this study. The number of selected variables and LVs were 300 and 6, resPectively. The correlation of determination in validation set （R^2） and root mean square error of prediction （RMSEP） by GA-PIS were 0.9900 and 4.4440, respectively.
出处
《中国农机化》
北大核心
2012年第2期132-135,共4页
Chinese Agricul Tural Mechanization
基金
国家自然科学基金资助项目(30760101
30972052)
新世纪优秀人才支持计划资助项目(NCET-09-0168)
江西省科技支撑计划项目(2009BNB05705)
江西省青年科学家(井岗之星)培养
江西省教育厅科学技术研究项目(GJJ08513)
