CAJ | 학술논문

为了对玉米秸秆的饲用品质进行可靠、便捷、快速的分析和评价,该研究以不同品种、密度、氮肥和水分处理的不同发育时期和不同部位玉米秸秆为试验材料,应用近红外光谱(NIRS)技术和偏最小二乘法(PLS),采用一阶导数+中心化+多元散射校正的光谱数据预处理方法,构建了玉米秸秆体外干物质消化率(IVDMD)、酸性洗涤纤维(ADF)、中性洗涤纤维(NDF)和可溶性糖(WSC)含量的NIRS分析模型.所建立的IVDMD、ADF、NDF和WSC含量的NIRS校正模型决定系数(R_(cal)~2)分别为0.9906、0.9870、O.9931和0.9802,交叉验证决定系数(R_(cv)~2)分别为0.9593、0.9413、0.9678和0.9342,外部验证决定系数(R_(val)~2)分别为0.9549、0.9353、0.9519和0.9191,各项标准差(SEC、SECV和SEP)为0.935~1.904,相对分析误差(RPD)均大于3.结果表明,各参数的NIRS分析模型可用于玉米秸秆饲用品质的分析和品种选育的快速鉴定.
위료대옥미갈간적사용품질진행가고、편첩、쾌속적분석화평개,해연구이불동품충、밀도、담비화수분처리적불동발육시기화불동부위옥미갈간위시험재료,응용근홍외광보(NIRS)기술화편최소이승법(PLS),채용일계도수+중심화+다원산사교정적광보수거예처리방법,구건료옥미갈간체외간물질소화솔(IVDMD)、산성세조섬유(ADF)、중성세조섬유(NDF)화가용성당(WSC)함량적NIRS분석모형.소건립적IVDMD、ADF、NDF화WSC함량적NIRS교정모형결정계수(R_(cal)~2)분별위0.9906、0.9870、O.9931화0.9802,교차험증결정계수(R_(cv)~2)분별위0.9593、0.9413、0.9678화0.9342,외부험증결정계수(R_(val)~2)분별위0.9549、0.9353、0.9519화0.9191,각항표준차(SEC、SECV화SEP)위0.935~1.904,상대분석오차(RPD)균대우3.결과표명,각삼수적NIRS분석모형가용우옥미갈간사용품질적분석화품충선육적쾌속감정.
In order to reliably, conveniently and rapidly analyze and evaluate forage quality of maize stover, the samples of maize stover from different varieties and treatments of density, nitrogenous fertilizer and water were used to establish near infrared reflectance spectroscopy (NIRS) calibration models of in vitro dry matter digestion (IVDMD), acid detergent fiber (ADF), neutral detergent fiber (NDF) and water soluble carbohydrate (WSC) of maize stover with near infrared reflectance spectroscopy (NIRS) technique, partial least square regression (PLS) and data pretreatment of Ist derivative+mean center+Multiple scarer correction. The results showed that determination coefficients of calibration (R_(cal)~2) about those models were 0.9906, 0.9870, 0.9931 and 0.9802 and those of cross validation (R_(cv)~2) and validation (R_(val)~2) were 0.9593(0.9549),0.9413(0.9353), 0.9678(0.9519) and 0.9342(0.9191) for IVDMD, ADF, NDF and WSC, respectively. Standard error of calibration, cross validation and prediction (SEC, SECV and SEP) ranged from 0.935 to 1.904. All values of relative percent differences (RPD) were greater than three. It demonstrated that these calibration models could be used to rapidly and accurately predict forage quality of maize stover and screen various samples in maize breeding.