【作者】 牛智有；

【导师】 韩鲁佳；

【作者基本信息】 中国农业大学 ， 农产品加工及贮藏工程， 2005， 博士

【摘要】 饲料化学品质的检测是保证饲料产品质量与安全的重要技术手段,为了研究探讨鱼粉和精料补充料的化学成分及其中肉骨粉含量的近红外反射光谱（NIRS）快速检测方法,收集了145个鱼粉、151个精料补充料样本和34个肉骨粉样本,在鱼粉和精料补充料中加入不同比例（鱼粉5～60%;精料补充料0.5～7%）的肉骨粉,在10000cm^-1～4000cm^-1谱区范围内,对试验样本进行了近红外反射光谱扫描。采用偏最小二乘（PLS）法,研究建立了鱼粉化学成分和氨基酸、精料补充料化学成分和总能、鱼粉和精料补充料中肉骨粉含量的近红外反射光谱定量分析模型,研究结果如下: 1.鱼粉中水分、粗蛋白、粗脂肪、粗灰分、总磷的化学分析值与NIRS定标模型预测值之间的决定系数R²达到0.90以上,相对标准差RSD均小于10%,相对分析误差RPD大于3;食盐的决定系数R²为0.8782,相对标准差RSD为9.86%,相对分析误差RPD为2.98;钙的决定系数R²为0.7575,相对标准差RSD为17.81%,相对分析误差RPD为2.14。结果表明,利用近红外光谱分析技术能够准确地检测鱼粉中水分、粗蛋白、粗脂肪、粗灰分、总磷的含量;食盐只能进行粗略估测;钙难于进行实际检测。 2.鱼粉中天冬氨酸、蛋氨酸、赖氨酸、苏氨酸、谷氨酸、甘氨酸、丙氨酸、缬氨酸、异亮氨酸、亮氨酸、苯丙氨酸、精氨酸、脯氨酸和总氨基酸的化学分析值与NIRS预测值的决定系数R²都达到0.87以上,相对标准差RSD均小于10%,相对分析误差RPD均大于3;酪氨酸的相关系数R²为0.8678,相对标准差RSD为8.65,相对分析误差RPD为2.77;组氨酸、丝氨酸和半胱氨酸的决定系数R²分别为0.9005、0.7436和0.3541,相对标准差RSD分别为14.19%、17.85%和33.85%;相对分析误差RPD分别为2.96、1.98和1.04。结果表明,利用近红外光谱分析技术能够较准确地检测鱼粉中天冬氨酸、蛋氨酸、赖氨酸、苏氨酸和总氨基酸等14种氨基酸的含量;酪氨酸只能进行粗略估测;组氨酸、丝氨酸和半胱氨酸难于进行实际检测。 3.精料补充料中水分、粗蛋白、粗脂肪、粗灰分、粗纤维、总磷、食盐和总能的化学分析值与NIRS定标模型预测值之间的决定系数R²均到0.90以上,相对标准差RSD均小于10%,相对分析误差RPD均大于3。结果表明,近红外光谱分析技术可以定量检测精料补充料中水分、粗蛋白、粗脂肪、粗灰分、粗纤维、总磷、食盐和总能的含量,并可获得理想的检测精度。 4.鱼粉中肉骨粉的NIRS定量分析模型,定标集真值与NIRS定标模型预测值之间的决定系数R²和标准差分别为0.9509和3.222;相对标准差RSD和相对分析误差RPD分别为9.25和4.79。验证集真值与NIRS预测值的决定系数r²以及标准差RMSEP分别为0.9668和2.68;相对标准差RSD和相对分析误差RPD分别为9.325和5.484。结果表明,用近红外光谱分析技术定量检测鱼粉中肉骨粉含量是可行的,并获得理想的预测精度。 5.精料补充料中肉骨粉含量的NIRS定量分析模型,定标集真值与NIRS定标模型预测值之间的决定系数R²和标准差分别为0.9447和0.392;相对标准差RSD和相对分析误差RPD分别为9.63和4.73。验证集真值与NIRS预测值决定系数r²以及标准差RMSEP分别为0.9618和0.345;相对标准差RSD和相对分析误差RPD分别为9.66和5.04。结果表明,用近红外光谱分析技术检测精料补充料中肉骨粉含量是可行的,并获得理想的预测精度。更多还原

【Abstract】 The measurement of feed quality is a crucial technical method to ensure the quality and safety of feed products. In order to investigate the method of near infrared reflectance spectroscopy （NIRS） technology for rapidly analysis of the chemical composition and meat and bone（MBM） content in fishmeal and concentrate supplement, 145 fishmeal samples, 151 concentrate supplement samples and 34 meat and bone meal（MBM） samples were collected, and deliberately adulterated with meat and bone meal （MBM） in the different proportion （fishmeal at 5-60%, concentrate supplement at 0.5-7%）. the samples were scanned at the NIRS region 1000 cm^-14000cm（-1）. The calibration models to predict the chemical conposition and amino acid in fishmeal、 the chemical conposition and gross energy、 the MBM content in fishmeal and concentrate supplement were developed using partial least squares （PLS） technique. The results were as follows:1 .The coefficient of determination in calibration （R²） of moisture, crude protein, crude fat, crude ash, total phosphor in fishmeal were all over 0.90, the RSD all less than 10%, and the RPD all over 3; the R² of NaCl was 0.8782, the RSD and RPD of salt were 9.86% and 2.98 respectively; the R² of Calcium was 0.7575, the RSD and RPD of calcium were 17.81% and 2.14 respectively. The results shows that NIRS analysis technique could be adopted to correctly measure the content of moisture, crude protein, crude fat, crude ash and total phosphor in fishmeal, and to predict salt content roughly. However, it could hardly be employed on measurement of Calcium content in practice.2.The coefficient of determination in calibration （R² ） of chemical analysis value and NIRS prediction value of aspartic acid, methionine, lysine, threonine, glutamic acid, glycine, alanine, valine, isoleucine, leucine, phenylalanine, arginine, proline and total amino acid were all over 0.87, the RSD all less than 10%, RPD all over 3; the R² of Tyrosine was 0.8678, the RSD was 8.65, the RPD was 2.77; the R² for histidine, serine and cysteine were 0.9005, 0.7436 and 0.3541 respectively, the RSD were 14.19%, 17.859% and 33.85% respectively, the RPD were 2.96, 1.98 and 1.04 respectively. The results shows that NIRS analysis technique could be used to measure the contents of 14 kinds of amino acids, such as aspartic acid, methionine, lysine, threonine and total amino acid in fish meal accurately and to predict Tyrosine content roughly. However, it can hardly be employed to measure histidine, serine and cysteine content in practice.3. The coefficient of determination in calibration （R²） of moisture, crude protein, crude fat, crude ash, total phosphor, salt, crude fiber and gross energy in concentrate supplement were all over 0.90, the RSD all less than 10%, the RPD all over 3. The results shows that NIRS analysis technique could be used to quantitatively predict the contents of moisture, crude protein, crude fat, crude ash, crude fiber, total phosphor, salt, and gross energy in concentrate supplement, and a good measuring precision could be expected.4.The NIRS quantitative analysis model was developed for measuring meat and bone meal（MBM） in fish meal. The coefficient of determination （R²） and standard errors （RMSEC） in calibration sets were0.9509 and 3.222 respectively; the RSD and the RPD were 9.25 and 4.79 respectively. The coefficient of determination （r^） and standard errors （RMSEP） in validation sets were 0.9668 and 2.68 respectively; the RSD and the RPD were 9.325 and 5.484 respectively. The results shows that it is feasible to employ NIRS analysis technique to quantitatively measure the MBM content in fish meal with a good measuring precision.5.The NIRS quantitative analysis model was developed for measuring meat and bone meal content in concentrate supplement. The coefficient of determination R² and standard errors （RMSEC） in calibration sets were 0.9447 and 0.392 respectively; the RSD and RPD 9.63 and 4.73 respectively. The coefficient of determination r² and standard errors （RMSEP） in validation sets were 0.9618 and 0.345 respect更多还原