CAJ | 학술논문

为了克服传统化学检测方法的缺陷，研究了不同氮营养下番茄叶片的近红外光谱特征，利用偏最小二乘回归算法建立了光谱数据与植物生化组分信息的定量分析模型，并采用决定系数 R2、交叉校验定标标准差 RM-SECV、相对分析误差 RPD和预测标准差RMSEP 验证了模型的优劣。实验表明，所建立的水分、全氮 NIR 模型的R2分别达到92．35％、86．15％，RMSECV 分别为0．346、0．129，RPD 分别为3．62、2．69，RMSEP 分别为0．209、0.111，预测精度满足实际的测定需求，说明运用近红外漫反射光谱监测番茄植株的生长状况是可行的。
위료극복전통화학검측방법적결함，연구료불동담영양하번가협편적근홍외광보특정，이용편최소이승회귀산법건립료광보수거여식물생화조분신식적정량분석모형，병채용결정계수 R2、교차교험정표표준차 RM-SECV、상대분석오차 RPD화예측표준차RMSEP 험증료모형적우렬。실험표명，소건립적수분、전담 NIR 모형적R2분별체도92．35％、86．15％，RMSECV 분별위0．346、0．129，RPD 분별위3．62、2．69，RMSEP 분별위0．209、0.111，예측정도만족실제적측정수구，설명운용근홍외만반사광보감측번가식주적생장상황시가행적。
In order to overcome the defects of traditional chemical determination method , the spectral characteristics in near-infrared region of tomato samples under different nitrogen nutrition were studied .Partial least-squares regression al-gorithm was applied to establish models for quantitative analysis between plant biochemical composition information and near-infrared spectral data , while determination coefficient ( R2 ) , root mean square error of cross validation ( RM-SECV), relative percent deviation (RPD) and root mean square error of prediction (RMSEP) were used as four main parameters to evaluate the performance of calibration models .Experiments showed that R 2 of NIR prediction models built for moisture and total nitrogen reached 92 .35%and 86 .15%respectively , RMSECV were 0 .346 and 0 .129 , RPD were 3.62 and 2.69, RMSEP were 0.209 and 0.111, meeting the practical requirement for content determination .That is, near infrared spectroscopy analysis is a feasible way to monitor growth of tomato .