基于RSM算法的烟叶含水率监测  

Tobacco Moisture Content Monitoring Based on RSM Algorithm

作　　者：尚晓明[1] 张娟利[1] 虎良词 SHANG Xiao-ming;ZHANG Juan-li;HU Liang-ci(Xingyi Normal University for Nationalities,Xingyi Guizhou 562400,China)

机构地区：[1]兴义民族师范学院,贵州兴义562400

出　　处：《林业机械与木工设备》2024年第4期21-26,共6页Forestry Machinery & Woodworking Equipment

基　　金：贵州省普通高等学校青年科技人才成长项目(黔教合KY字[2020]213);黔西南州科技计划(2019-2-54)。

摘　　要：为探讨将随机子空间RSM(Random Subspace Method)算法应用于烤烟烟叶含水率的监测中,采摘贵烟8号烟叶样本,在09:00-12:00时间段进行可见光采样,并对图像亮度进行梯度处理,以此模拟全天光线变化。采用烟叶样本实测含水率和图像RGB三阶颜色矩数据作为数据集,对样本使用RSM算法建立含水率回归模型,并与LM(Levenberg Marquardt)神经网络算法和支持向量机(Support Vector Machine,SVM)算法进行比较。结果表明,基于烟叶RGB颜色矩的RSM算法具有较好的应用效果,其回归模型决定系数R 2为0.9202,均方根误差(RMSE)为0.56%,相对分析误差(RPD)为3.5483。故基于随机子空间RSM算法的烟叶含水率回归模型具有较好的稳定性,能实现对烟叶含水率的监测。In order to explore the application of RSM algorithm in the monitoring of tobacco moisture content,the samples of tobacco leaf No.8 of Guiyan were collected,and the visible light sampling was carried out from 09:00 to 12:00,and the gradient processing of image brightness was carried out to simulate the change of all-day light.The measured moisture content of tobacco leaf samples and the RGB third-order color moment data of the image were used as data sets.The RSM algorithm was used to establish the moisture content regression model for the samples,and compared with the LM neural network algorithm and the SVM algorithm.The results showed that the RSM algorithm based on the RGB color moments of tobacco leaves had good application effect.The determination coefficient of the regression model was 0.9202,the root mean square error(RMSE)was 0.56%,and the relative analysis error(RPD)was 3.5483.Therefore,the regression model of leaf moisture content based on the random subspace RSM algorithm has good stability and can realize the monitoring of tobacco moisture content.

关 键 词：烟叶含水率 RSM算法 RGB颜色矩 监测 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置] S776[自动化与计算机技术—控制科学与工程]