无损检测5种药用植物叶片含水量的通用模型研究

Study of Generic Model to Nondestructively Test Leaf Moisture Content in Five Medicinal Plants

关键词:  电容  电阻  叶厚  叶片含水量  模型

Key words:  capacitance  resistance  thickness  leaf moisture content  model

摘要:

目的 利用电测法对药用植物叶片含水量进行快速、准确和无损检测。方法 以女贞、何首乌、银杏、葛和龙葵叶片为研究对象，自行设计平行板电容传感器，改进电阻测量方法，对叶片电容、电阻和叶厚进行检测。采用SPSS 19.0软件对测量数据进行组内相关系数分析，验证数据的可靠性。将叶片分成训练集和测试集，用Excel对训练集进行回归分析，建立叶片含水量与电容、电阻和叶厚的拟合方程，并利用拟合方程对测试集叶片含水量进行预测。结果 相同药用植物不同叶片间的电容测量值可靠性良好，何首乌、葛和龙葵叶片电阻测量值可靠性良好，女贞和银杏电阻测量值可靠性一般。不同药用植物叶片间的电容和电阻测量值可靠性一般，组间叶厚测量值可靠性良好。经Excel线性回归，R²为0.959 7，调整R²为0.951 0，显著性值P=5.36×10^-10，拟合方程为Y=23.548 3+0.021 6 X₁+12.705 8 X₂+106.786 1 X₃，DW=2.284，模型拟合效果良好。利用该模型对测试集叶片含水量进行预测，与烘干法比较误差值在1.98%~-1.55%之间。结论 该模型可以作为该5种药用植物叶片含水量预测的通用模型。

Abstract:

OBJECTIVE To rapidly, correctly and nondestructively detect leaf moisture content by electrical measuring method in medicinal plants. METHODS The leaf capacitance, resistance and thickness were detected by using the self-designed parallel-plate capacitor and improving resistance measuring method in privet, polygonum multiflorum, ginkgo, kudzu and solanum sigrum sinne. Measured data was analyzed by SPSS 19.0 software for intraclass correlation coefficient verify the reliability of the data. The leaves were divided into training set and test set. The training set was analyzed with Excel regression. The fitting equation was established among leaf moisture content, capacitance, resistance and thickness. The fitting equation was used to predict the leaf moisture content in the test set. RESULTS The data reliability of capacitance among different leaves of the same medicinal plant was good. The data reliability of resistance among different leaves in polygonum multiflorum, kudzu and solanum sigrum sinne was good, and the data reliability of leaf resistance was general in privet and Ginkgo. The data reliability of capacitance and resistance among different medicinal plants was general. The data reliability of thickness among groups of different medicinal plants was good. By Excel regression analysis, R² was 0.959 7, adjusted R² was 0.951 0, significant value P=5.36×10^-10, fitting equation Y=23.548 3+0.021 6 X₁+12.705 8 X₂+106.786 1 X₃, DW=2.284, the fitting effect of the model was good. The model was used to predict the moisture content of the test set, and the errors were between 1.98% and -1.55% compared with the drying method. CONCLUSION The model can be used as a generic model for predicting leaf water content of the five medicinal plants.