基于GA-LS-SVM的水稻叶片含氮率预测  

Prediction of nitrogen content rate of paddy rice leaf based on GA-LS-SVM

作　　者：孙俊[1,2] 毛罕平[1] 羊一清[1] 

机构地区：[1]江苏大学江苏省现代农业装备与技术重点实验室,江苏镇江212013 [2]江苏大学电气信息工程学院,江苏镇江212013

出　　处：《江苏大学学报（自然科学版）》2010年第1期6-10,共5页Journal of Jiangsu University：Natural Science Edition

基　　金：国家高技术研究发展计划"863"项目(2008AA10Z204);中国博士后科学基金面上资助项目(20070420972);江苏大学高级专业人才科研启动基金资助项目(05JDG050);常州市青年科技人才培养计划项目(CQ2008009)

摘　　要：采用水稻无土栽培方法人为控制含氮率,在水稻某特定生长期,同时测量水稻冠层反射率和叶片含氮率,建立了基于冠层反射率的水稻叶片氮含率的回归预测模型.通过分析不同氮环境下各冠层反射率光谱图,确定了与水稻含氮率相关性高的特征波段.针对最小二乘支持向量机(leastsquares support vector machines,LS-SVM)参数难定问题,采用遗传算法对LS-SVM参数进行优化.试验结果表明,传统人为选定参数的LS-SVM算法模型的平均回判精确率达到97.21%,预测平均误差率达到5.70%,遗传算法最小二乘支持向量机(genetic algorithm least squares support vec-tor machines,GA-LS-SVM)算法模型的平均回判精确率达到99.60%,预测平均误差率达到2.72%.GA-LS-SVM算法模型的回判及预测效果均明显优于人为选定参数的LS-SVM算法.A prediction model of paddy rice leaf nitrogen content rate was built based on canopy spectrum reflectivity. The paddy rice of every nitrogen content was cultivated by man-made control, At a certain vegetation period, the paddy rice canopy spectrum reflectivity was gathered and the leaf nitrogen content rate was measured at the same time. Each canopy spectrum image was analyzed, and the characteristic wave band was chosen which matches the high relativity coefficient. Because the LS -SVM＇s parameters were difficult to be confirmed, genetic algorithm was adopted to carry out an optimization on LS - SVM parameter and the GA - LS - SVM algorithm was composed. Examination results indicate that, the traditional LS- SVM model＇s average return-judge accuracy reaches 97.21% , and it＇s average error ratio of prediction reaches 5.70%. The GA - LS - SVM model＇s average return-judge accuracy reaches 99.60% , and it＇s average error ratio of prediction reaches 2.72%. So the GA - LS - SVM model＇s return-judge accuracy and the average error ratio are better than those of LS - SVM algorithm model whose parameters are set artificially.

关 键 词：水稻 氮素 冠层 光谱反射率 GA—LS—SVM算法 

分 类 号：Q945.17[生物学—植物学]