1 材料与方法
1.1 饲料原料近红外光谱的采集与筛选
1.2 常规化学指标测定
1.3 有效能值测定
1.4 样品分集
1.5 定标过程和交互验证
1.6 模型的外部验证
2 结果与分析
2.1 9种饲料原料生长猪消化能和代谢能
表1 9种饲料原料生长猪消化能和代谢能的计算方程(干物质基础)Table 1 Equations used to calculate digestible energy and metabolizable energy of 9 feed ingredients for growing pigs (DM basis) MJ/kg |
| 饲料原料Feed ingredients | 消化能DE | 代谢能ME |
|---|---|---|
| 全脂米糠Full-fat rice bran | DE=11.52+0.23×EE | ME=1.075+0.890×DE |
| 大麦Barley | DE=16.22-0.25×ADF | ME=15.90-0.24×ADF |
| 麦麸Wheat bran | DE=18.26-0.807×Ash | ME=17.302-0.715×Ash |
| 玉米蛋白粉Corn gluten meal | DE=16.24+0.058×CP | ME=15.963+0.057×CP |
| 花生粕Peanut meal | DE=2.866+0.225×CP | ME=17.78-0.182×NDF |
| 棉籽粕Cottonseed meal | DE=11.876+0.052×CP-0.228×CF | ME=14.117-0.253×CF |
| 菜籽粕Rapeseed meal | DE=14.284+0.053×CP+0.502×EE-0.199×ADF | ME=0.319+0.894×DE |
| 次粉Wheat shorts | DE=22.289-2.232×Ash | ME=21.469-2.206×Ash |
| 玉米胚芽粕Corn germ meal | DE=14.883-0.169×ADF | ME=14.171-0.161×ADF |
EE:粗脂肪 ether extract;ADF:酸性洗涤纤维 acid detergent fiber;Ash:粗灰分;CP:粗蛋白质 crude protein;NDF:中性洗涤纤维 neutral detergent fiber;CF:粗纤维 crude fiber。 |
表2 9种饲料原料生长猪消化能和代谢能(干物质基础)Table 2 Digestible energy and metabolizable energy of 9 feed ingredients for growing pigs (DM basis) MJ/kg |
| 饲料原料 Feed ingredients | 样品数 n | 有效能 Available energy | 最大值 Max | 最小值 Min | 平均值 Mean | 标准差 SD | 变异系数 CV/% |
|---|---|---|---|---|---|---|---|
| 全脂米糠 | 369 | 消化能 | 18.07 | 14.91 | 16.81 | 0.56 | 3.33 |
| Full-fat rice bran | 代谢能 | 17.58 | 14.24 | 16.24 | 0.59 | 3.64 | |
| 大麦 | 521 | 消化能 | 16.22 | 13.87 | 15.20 | 0.66 | 4.33 |
| Barley | 代谢能 | 15.90 | 13.64 | 14.92 | 0.63 | 4.53 | |
| 麦麸 | 174 | 消化能 | 15.35 | 11.60 | 13.08 | 0.59 | 4.50 |
| Wheat bran | 代谢能 | 15.10 | 11.39 | 12.82 | 0.56 | 4.33 | |
| 玉米蛋白粉 | 223 | 消化能 | 20.92 | 20.21 | 20.54 | 0.10 | 1.00 |
| Corn gluten meal | 代谢能 | 19.85 | 19.34 | 19.58 | 0.07 | 1.00 | |
| 花生粕 | 326 | 消化能 | 16.40 | 13.98 | 15.10 | 0.41 | 2.71 |
| Peanut meal | 代谢能 | 15.81 | 12.12 | 13.94 | 0.54 | 3.90 | |
| 棉籽粕 | 237 | 消化能 | 14.16 | 10.17 | 11.86 | 0.77 | 6.43 |
| Cottonseed meal | 代谢能 | 13.02 | 9.57 | 11.05 | 0.67 | 6.04 | |
| 菜籽粕 | 283 | 消化能 | 15.33 | 3.22 | 11.69 | 1.48 | 12.61 |
| Rapeseed meal | 代谢能 | 14.22 | 2.66 | 10.74 | 1.43 | 13.29 | |
| 次粉 | 138 | 消化能 | 17.19 | 11.44 | 14.87 | 0.67 | 4.48 |
| Wheat shorts | 代谢能 | 16.34 | 11.01 | 13.82 | 0.80 | 5.76 | |
| 玉米胚芽粕 | 160 | 消化能 | 14.03 | 12.16 | 12.84 | 0.43 | 3.39 |
| Corn germ meal | 代谢能 | 13.28 | 11.62 | 12.22 | 0.39 | 3.17 |
2.2 9种饲料原料的原始近红外光谱图和最佳有效能定标模型的预处理光谱图
图1 全脂米糠、大麦和麦麸的原始近红外光谱图和最佳有效能定标模型的预处理光谱图图A表示全脂米糠无预处理光谱;图B表示全脂米糠一阶导数结合矢量归一化处理光谱;图C表示大麦无预处理光谱;图D表示大麦多元散射校正处理光谱;图E表示麦麸无预处理光谱;图F表示麦麸一阶导数结合多元散射校正处理光谱;图G表示麦麸一阶导数结合矢量归一化处理光谱。 Fig.1 Original near-infrared spectra of full-fat rice bran, barley and wheat bran, and preprocessing spectra of optimal available energy calibration models Figure A represented the full-fat rice bran spectra pretreated with none pretreatment; figure B represented the full-fat rice bran spectra pretreated with first derivative combined with vector normalization; figure C represented the barley spectra pretreated with none pretreatment; figure D represented the barley spectra pretreated with multiplicative scatter correction; figure E represented the wheat bran spectra pretreated with none pretreatment; figure F represented the wheat bran spectra pretreated with first derivative combined with multiplicative scatter correction; figure G represented the wheat bran spectra pretreated with first derivative combined with vector normalization. |
图2 玉米蛋白粉、花生粕和棉籽粕的原始近红外光谱图和最佳有效能定标模型的预处理光谱图图A表示玉米蛋白粉无预处理光谱;图B表示玉米蛋白粉一阶导数结合矢量归一化处理光谱;图C表示花生粕无预处理光谱;图D表示花生粕一阶导数结合矢量归一化处理光谱;图E表示棉籽粕无预处理光谱;图F表示棉籽粕二阶导数处理光谱;图G表示棉籽粕一阶导数结合矢量归一化处理光谱。 Fig.2 Original near-infrared spectra of corn gluten meal, peanut meal and cottonseed meal, and preprocessing spectra of optimal available energy calibration models Figure A represented the corn gluten meal spectra pretreated with none pretreatment; figure B represented the corn gluten meal spectra pretreated with first derivative combined with vector normalization; figure C represented the peanut meal spectra pretreated with none pretreatment; figure D represented the peanut meal spectra pretreated with first derivative combined with vector normalization; figure E represented the cottonseed meal spectra pretreated with none pretreatment; figure F represented the cottonseed meal spectra pretreated with second derivative; figure G represented the cottonseed meal spectra pretreated with first derivative combined with vector normalization. |
图3 次粉、玉米胚芽粕和菜籽粕的原始近红外光谱图和最佳有效能定标模型的预处理光谱图图A表示次粉无预处理光谱;图B表示次粉一阶导数结合多元散射校正处理光谱;图C表示玉米胚芽粕无预处理光谱;图D表示玉米胚芽粕一阶导数结合矢量归一化处理光谱;图E表示菜籽粕无预处理光谱;图F表示菜籽粕一阶导数结合矢量归一化处理光谱;图G表示菜籽粕一阶导数结合多元散射校正处理光谱。 Fig.3 Original near-infrared spectra of wheat shorts, corn germ meal and rapeseed meal, and preprocessing spectra of optimal available energy calibration models Figure A represented the wheat shorts spectra pretreated with none pretreatment; figure B represented the wheat shorts spectra pretreated with first derivative combined with multiplicative scatter correction; figure C represented the corn germ meal spectra pretreated with none pretreatment; figure D represented the corn germ meal spectra pretreated with first derivative combined with vector normalization; figure E represented the rapeseed meal spectra pretreated with none pretreatment; figure F represented the rapeseed meal spectra pretreated with first derivative combined with vector normalization; figure G represented the rapeseed meal spectra pretreated with first derivative combined with multiplicative scatter correction. |
2.3 9种饲料原料生长猪有效能定标模型的R2cv和RMSECV随模型因子数的变化
图4 全脂米糠、大麦和麦麸生长猪消化能和代谢能定标模型的交互验证决定系数和交互验证标准差随模型因子数变化的曲线R2cv表示交互验证决定系数;RMSECV表示交互验证标准差。图5和图6同。 Fig.4 Coefficient of determination of cross-external validation and root mean square error of cross-external validation curves for digestible energy and metabolizable energy calibration models of full-fat rice bran, barley and wheat bran for growing pigs varied with model factors R2cv means coefficient of determination of cross-external validation; RMSECV means root mean square error of cross-external validation. The same as Fig.5 and Fig.6. |
图5 玉米蛋白粉、花生粕和棉籽粕生长猪消化能和代谢能定标模型的交互验证决定系数和交互验证标准差随模型因子数变化的曲线Fig.5 Coefficient of determination of cross-external validation and root mean square error of cross-external validation curves for digestible energy and metabolizable energy calibration models of corn gluten meal, peanut meal and cottonseed meal for growing pigs varied with model factors |
图6 菜籽粕、次粉和玉米胚芽粕生长猪消化能和代谢能定标模型的交互验证决定系数和交互验证标准差随模型因子数变化的曲线Fig.6 Coefficient of determination of cross-external validation and root mean square error of cross-external validation curves for digestible energy and metabolizable energy calibration models of rapeseed meal, wheat shorts and corn germ meal for growing pigs varied with model factors |
2.4 9种饲料原料最佳生长猪消化能和代谢能定标模型
表3 9种饲料原料最佳生长猪消化能和代谢能定标模型Table 3 Optimal digestible energy and metabolizable energy calibration models of 9 feed ingredients for growing pigs |
| 饲料原料 Feed ingredients | 模型 Models | 维数 Factors | 光谱范围 Spectral regions/cm-1 | 预处理方法 Preprocessing methods | R2c1) | RMSEC2) | R2cv3) | RMSECV4) | RPDcv5) |
|---|---|---|---|---|---|---|---|---|---|
| 全脂米糠 | 消化能 | 9 | 8 454.9~7 498.4; 5 778.1~5 446.3 | 1st D+SNV | 0.99 | 0.06 | 0.98 | 0.07 | 7.79 |
| Full-fat rice bran | 代谢能 | 9 | 8 454.9~7 498.4; 6 102.1~5 770.3 | 1st D+SNV | 0.98 | 0.08 | 0.98 | 0.09 | 6.87 |
| 大麦 | 消化能 | 8 | 9 049.0~7 498.4; 6 102.1~5 446.3 | MSC | 0.96 | 0.08 | 0.95 | 0.08 | 4.59 |
| Barley | 代谢能 | 8 | 9 049.0~7 498.4; 6 102.1~5 446.3 | MSC | 0.96 | 0.07 | 0.95 | 0.09 | 4.65 |
| 麦麸 | 消化能 | 13 | 9 403.7~4 242.8 | 1st D+MSC | 0.99 | 0.06 | 0.98 | 0.07 | 7.48 |
| Wheat bran | 代谢能 | 13 | 9 403.7~4 242.8 | 1st D+SNV | 0.99 | 0.07 | 0.98 | 0.08 | 6.45 |
| 玉米蛋白粉 | 消化能 | 14 | 9 403.8~6 094.3; 4 605.5~4 242.9 | 1st D+SNV | 0.96 | 0.02 | 0.93 | 0.03 | 3.81 |
| Corn gluten meal | 代谢能 | 9 | 7 506.1~5 446.3; 4 605.5~4 242.9 | 1st D+SNV | 0.94 | 0.02 | 0.91 | 0.02 | 3.42 |
| 花生粕 | 消化能 | 11 | 7 428.9~5 446.3; 4 605.5~4 242.9 | 1st D+SNV | 0.99 | 0.04 | 0.99 | 0.04 | 10.00 |
| Peanut meal | 代谢能 | 13 | 9 403.8~6 094.3; 5 454.0~4 597.7 | 1st D+SNV | 0.98 | 0.07 | 0.97 | 0.08 | 6.32 |
| 棉籽粕 | 消化能 | 8 | 8 910.1~5 446.3 | 2nd D | 0.98 | 0.11 | 0.97 | 0.13 | 6.05 |
| Cottonseed meal | 代谢能 | 11 | 8 910.1~6 094.3 | 1st D+SNV | 0.99 | 0.07 | 0.99 | 0.08 | 8.38 |
| 菜籽粕 | 消化能 | 11 | 9 403.8~5 446.3 | 1st D+SNV | 0.99 | 0.11 | 0.98 | 0.12 | 6.91 |
| Rapeseed meal | 代谢能 | 12 | 9 403.8~7 498.3; 6 102.0~4 242.9 | 1st D+MSC | 0.99 | 0.14 | 0.98 | 0.16 | 7.62 |
| 次粉 | 消化能 | 10 | 9 403.8~5 446.3 | 1st D+MSC | 0.98 | 0.08 | 0.98 | 0.51 | 6.34 |
| Wheat shorts | 代谢能 | 11 | 9 403.8~5 446.3 | 1st D+MSC | 0.99 | 0.08 | 0.97 | 0.11 | 5.84 |
| 玉米胚芽粕 | 消化能 | 7 | 8 454.9~7 498.3; 4 605.5~4 242.9 | 1st D+SNV | 0.98 | 0.03 | 0.97 | 0.04 | 6.23 |
| Corn germ meal | 代谢能 | 7 | 8 454.9~7 498.3; 4 605.5~4 242.9 | 1st D+SNV | 0.97 | 0.03 | 0.97 | 0.04 | 5.47 |
1)定标决定系数。Coefficient of determination of calibration. | |
2)定标标准差。Root mean square error of calibration. | |
3)交互验证决定系数。Coefficient of determination of cross-external validation. | |
4)交互验证标准差。Root mean square error of cross-external validation. | |
5)交互验证相对分析误差。Residual predictive deviation of cross-external validation. |
2.5 9种饲料原料最佳生长猪消化能和代谢能定标模型的外部验证
表4 9种饲料原料最佳生长猪消化能和代谢能定标模型的外部验证结果(干物质基础)Table 4 Results of external validation of optimal digestible energy and metabolizable energy calibration models of 9 feed ingredients for growing pigs (DM basis) |
| 饲料原料 Feed ingredients | 模型 Models | 样品数 n | 平均值 Mean/ (MJ/kg) | 标准差 SD/ (MJ/kg) | R2v1) | RMSEP2) | RPDv3) |
|---|---|---|---|---|---|---|---|
| 全脂米糠 | 消化能 | 11 | 16.81 | 0.73 | 0.85 | 1.30 | 2.13 |
| Full-fat rice bran | 代谢能 | 11 | 16.11 | 0.62 | 0.72 | 1.27 | 2.00 |
| 大麦 | 消化能 | 11 | 14.71 | 0.34 | 0.88 | 0.19 | 1.73 |
| Barley | 代谢能 | 11 | 13.37 | 0.48 | 0.83 | 0.20 | 2.28 |
| 麦麸 | 消化能 | 10 | 13.18 | 0.59 | 0.92 | 0.21 | 2.80 |
| Wheat bran | 代谢能 | 10 | 12.81 | 0.51 | 0.87 | 0.27 | 1.91 |
| 玉米蛋白粉 | 消化能 | 12 | 20.48 | 0.49 | 0.80 | 0.27 | 1.84 |
| Corn gluten meal | 代谢能 | 12 | 19.30 | 0.70 | 0.71 | 0.46 | 1.24 |
| 花生粕 | 消化能 | 12 | 15.54 | 0.43 | 0.79 | 0.25 | 2.07 |
| Peanut meal | 代谢能 | 12 | 13.71 | 0.49 | 0.64 | 0.33 | 1.67 |
| 棉籽粕 | 消化能 | 12 | 12.69 | 0.74 | 0.72 | 0.41 | 1.62 |
| Cottonseed meal | 代谢能 | 12 | 11.37 | 0.46 | 0.62 | 0.41 | 1.11 |
| 菜籽粕 | 消化能 | 12 | 13.14 | 1.05 | 0.84 | 0.44 | 2.53 |
| Rapeseed meal | 代谢能 | 12 | 12.04 | 1.23 | 0.81 | 0.55 | 2.21 |
| 次粉 | 消化能 | 7 | 14.72 | 0.70 | 0.82 | 0.36 | 1.91 |
| Wheat shorts | 代谢能 | 7 | 13.99 | 0.95 | 0.62 | 0.62 | 1.52 |
| 玉米胚芽粕 | 消化能 | 10 | 12.58 | 0.32 | 0.73 | 0.18 | 1.80 |
| Corn germ meal | 代谢能 | 10 | 12.02 | 0.38 | 0.62 | 0.23 | 1.62 |
1)外部验证决定系数。Coefficient of determination of external validation. | |
2)外部验证标准差。Root mean square error of external validation. | |
3)外部验证相对分析误差。Residual predictive deviation of external validation. |

京公网安备 11010802041926号