近红外光谱结合ANN法快速测定水稻叶片氮含量  

Rapid Analysis of Rice Leaf Nitrogen Using Near Infrared Spectroscopy and Artificial Neural Network

作　　者：周萍[1] 张广才[1] 王佼[1] 周崇俊[1] 韩晓日[1] 

机构地区：[1]沈阳农业大学土地与环境学院,辽宁沈阳110866

出　　处：《黑龙江农业科学》2011年第4期22-25,共4页Heilongjiang Agricultural Sciences

基　　金：教育部留学回国人员科研启动基金资助项目

摘　　要：应用近红外(NIR)光谱和误差反传人工神经网络(BP-ANN)方法建立了水稻叶片氮素含量的定量分析模型。首先对近红外光谱进行Savitzky-Golay求导处理,然后通过相关系数法选择波长范围,采用偏最小二乘回归PLS降维并输入BP-ANN建立校正模型,用验证样品对校正模型进行验证。结果表明:BP-ANN最佳模型的预测相关系数(RP)为0.974 7,预测标准误差(SEP)为4.005,预测相对标准差(RPD)为3.109。表明BP-ANN模型稳健可靠,可较好地用于水稻叶片氮素的快速测定。The models of quantitative analysis of nitrogen in the rice leaf were established by using near infrared spectroscopy（NIS）coupled with the back propagation-artificial neural network method（BP-ANN）.Firstly,the data of original spectra were pretreated by Savitzky-Golay derivative.Secondly,the wavelength range of model was optimized by using correlation coefficient method.Finally,PLS dimension-reducing was input into BP-ANN.The calibration models were established by calibration set and validated by prediction set.The results showed that the related coefficient（RP）of the best prediction for nitrogen was 0.974 7,the standard errors of prediction（SEP）for nitrogen was 4.005,and ratio of performance deviation（RPD）was 3.109.Therefore,the method could be applied to fast and accurate determination of nitrogen in the rice leaf.

关 键 词：人工神经网络 近红外光谱 水稻叶片 氮素 

分 类 号：S511[农业科学—作物学]