基于介电特性的稻谷干燥含水率检测模型构建与验证

Construction and Verification of Detection Model of Moisture Content for Paddy Rice Drying Based on Dielectric Properties

为建立稻谷通风干燥含水率在线监测系统,利用自制可替换平行极板电容器与阻抗分析仪构建系统测定模型,以辽-盐粳98为试验物料,测定不同极板材料、极板类型、极板绝缘性及通风网板网格尺寸条件下稻谷含水率与电容值间的数值关系。首先探究极板材料与结构参数对电容值的影响,同时测定1×10^(3)~1×10^(6)Hz频率段内稻谷的介电常数ε’与介质损耗因数ε"。采用基于X-Y共生距离(SPXY)算法划分样本集,对60个不同含水率稻谷样本按照比例2∶1划分为校正集和预测集,并利用无信息变量消除法(UVE)、UVE-SPA(连续投影法)联合筛选出ε’、ε"以及ε’和ε"结合变量的特征变量数及所选频率点,再以UVE、SPA和UVE-SPA选取方法获得的频率数据作为因变量建立特征变量与含水率间支持向量回归(SVR)模型,最终对该支持向量机回归模型进行验证。结果表明:选用通风钢丝网作极板材料、网格尺寸为1cm进行试验,含水率测定效果最佳;介电常数随稻谷含水率增加而增加,随频率增加先增加再降低;全面考虑ε’和ε"两个因变量,经UVE-SPA联合筛选特征变量建立的支持向量机回归模型预测精度最优,预测系数达0.9795,预测均方根误差为0.01%,验证了该测量装置及支持向量机回归模型的可靠性,为实现稻谷通风干燥过程水分快速无损检测提供基础数值参考。

To establish an online monitoring system for the moisture content of paddy rice during air drying,a system measurement model was established with a self-made replaceable parallel plate capacitor and impedance analyzer.Using Liao-Yanjing 98 as the test material,the numerical relationship between moisture content and capacitance of paddy rice was determined under different plate materials,plate types,plate insulation,and grid size of ventilation mesh.In this study,the influence of plate material and structural parameters on the capacitance value was explored,and the dielectric constant ε’and dielectric loss factor ε"of rice in the frequency range of 1×10^(3)-1×10^(6)Hz were measured.The X-Y co-incidence distance(SPXY)algorithm was used to divide the sample set,and 60 samples with different moisture content were divided into a calibration set and prediction set according to the ratio of 2∶1.The Uninformative Variable Elimination(UVE)and UVE-SPA(Successive Projection Algorithm)were used to screen out the number of characteristic variables and selected frequency points of ε’,ε"and combined variables of ε’and ε".Then,the frequency data obtained by UVE,SPA,and UVE-SPA selection methods were used as dependent variables to establish and verify the Support Vector Regression(SVR)model between characteristic variables and water content.The results showed that it performance best when the ventilated steel wire mesh is selected as the plate material and the mesh size is 1cm for the test of moisture content;the dielectric constant increases with the increase of rice moisture content,and first increases and then decreases with the increase of frequency.Comprehensively consider the two dependent variablesε’andε",the support vector machine regression model established with characteristic variables joint screened by UVE-SPA has the best prediction accuracy,with a prediction coefficient of 0.9795,and a prediction root mean square error of 0.01%,which verifies the reliability of the measuring device and the support vector machine regression model.This study provides a basic numerical reference for the rapid and non-destructive detection of moisture content during the ventilating and drying process of paddy rice.