摘要
采用两种不同的波长变量选择方法对南疆冬枣进行偏最小二乘(PLS)建模分析,使用连续投影算法建立的模型预测标准偏差(RMSEP)为1.0672,校正标准偏差(RMSEC)为0.5997,相关系数(R)为0.9017,主因子数为10。使用竞争性自适应加权采样算法建立的模型预测标准偏差(RMSEP)为1.0941,校正标准偏差(RMSEC)为0.6148,相关系数(R)为0.9275,主因子数为7。结果表明连续投影算法和竞争性自适应加权采样算法都能够有效地优选出全光谱的256个变量中的13个变量,降低建模的波长变量,减小了模型的复杂性,提高了模型的精度。两种算法在筛选南疆冬枣的特征波长是可行的。
Partial least squares(PLS)modeling analysis was carried out for winter jujube in southern xinjiang by adopting two different wavelength variable selection methods.The standard deviation(RMSEP)of the model established by the continuous projection algorithm was 1.0672,the correct standard deviation(RMSEC)was 0.5997,the correlation coefficient(R)was 0.9017,and the number of main factors was 10.The standard deviation(RMSEP),the corrected standard deviation(RMSEC),the correlation coefficient(R)and the main factor number were 0.9275.The results show that both the continuous projection algorithm and the competitive adaptive weighted sampling algorithm can effectively select 13 of the 256 variables in the full spectrum,reduce the wavelength variables,reduce the complexity of the model and improve the accuracy of the model.The two algorithms are feasible to screen the characteristic wavelengths of southern Xinjiang winter jujube.
作者
李伟
罗华平
索玉婷
陈冲
LI Wei;LUO Hua-ping;SUO Yu-ting;CHEN Chong(College of Mechanical and Electrical Engineering,Tarim University,Alar 843300,Xinjiang,China;The Key Laboratory of Colleges and Universities under the Department of Education of Xinjiang Uygur Autonomous Region)
出处
《新疆农机化》
2019年第5期20-23,共4页
Xinjiang Agricultural Mechanization
基金
国家自然科学基金（11464039,11164023）
关键词
竞争性自适应加权
连续投影
南疆冬枣
Competitive adaptive weighted
Continuous projection
Winter jujube
