摘要
量化研究烘烤变黄期烟叶颜色变化,实现烘烤变黄期叶片含水量的无损检测。利用色差仪检测‘K326’中部叶不同变黄温度(烤前32,34,36,38,40,42℃)下的叶片颜色参数(L*、a*、b*、C、H°),通过因子分析法对作为输入变量的颜色参数进行筛选,分别构建不同温度点的叶片含水量预测模型。网络模型训练结果表明:各温度点叶片含水量预测模型的模拟值与目标值的回归系数均达到显著水平或极显著水平,各网络模型预测误差在0~2.0范围内的样本占比达到88%以上,预测误差在0~3.0范围内的样本占比均达到95%以上。构建的不同变黄温度叶片含水量预测模型具有较高的准确性,在烘烤变黄期可利用BP神经网络基于烟叶颜色参数进行叶片含水量的快速无损估测。
Quantitative study on color change of flue cured tobacco leaves during yellowing period.Non-destructive testing of leaf moisture content during yellowing stage.The middle leaves of'K326'were used as experimental materials.The leaf color parameters(L*,a*,b*,C,H°)were measured colorimeter under different yellowing temperatures(32,34,36,38,40,42℃).The color parameters were screened by factor analysis.The prediction models of leaf moisture content at different temperature points were established.The network model training results showed that:the regression coefficients between the simulated value and the target value of each prediction model reached a significant level or extremely significant level.The proportion of samples with prediction error in the range of 0-2.0 was more than 88%.And the proportion of samples with prediction error in the range of 0-3.0 was more than 95%.The prediction model of leaf moisture content with different yellowing temperature had high accuracy.BP neural network could be used to estimate leaf moisture content rapidly and non-destructively based on tobacco color parameters during yellowing period.
作者
张广普
李峥
黄广华
徐志强
钟永健
赵程虹
ZHANG Guangpu;LI Zheng;HUANG Guanghua;XU Zhiqiang;ZHONG Yongjian;ZHAO Chenghong(China Tobacco Zhejiang Industrial Co.,Ltd.,Hangzhou,Zhejiang 310024,China)
出处
《天津农业科学》
CAS
2021年第10期62-67,73,共7页
Tianjin Agricultural Sciences
关键词
烤烟
变黄期
BP神经网络
颜色参数
叶片含水量
预测模型
flue-cured tobacco
yellowing stage
back propagation artificial neural network
color parameters
moisture content of tobacco leaves
Prediction model
