摘要
应用近红外透射光谱技术,采用3种不同回归统计分析方法建立精米蛋白质含量(PC)定量回归方程。结果表明,用改进的最小二乘法(MPLS)、偏最小二乘法(PLS)和主成分回归法(PCR)进行校正时,校正标准误差(SEC)、交叉检验标准误差(SECV)分别为0.1258、0.1340(MPLS),0.1177、0.1175(PLS),0.1207、0.1275(PCR)。校正相关系数(RSQ)和交叉验证相关系数(1-VR)分别为0.9941、0.9931(MPLS),0.9950、0.9942(PLS),0.9947、0.9942(PCR)。由此可见,3种回归统计方法在建立精米蛋白质含量回归方程时差异不明显,都具有较好的预测效果。近红外透射光谱法作为一种快速而准确的定量分析手段,在稻米加工企业品质管理、大米品质分析和大米贸易检测上有广阔的应用前景。
Three different regression statistic methods,namely Modified Partial Least Square(MPLS),Parˉtial LeastSquare(PLS),and Principal ComponentRegression(PCR)were examined to establish a quantitaˉtive regression equation for the determination of protein content in milled rice with near infrared transmitˉtance spectrometry.The results indicated that the standard errors for calibration(SEC)and cross validaˉtion(SECV)were0.1258and0.1340for MPLS,0.1177and0.1175for PLS,0.1207and0.1275for PCR,respectively;and the correlation coefficients for calibration(RSQ)and cross validation(1-VR)were0.9941and0.9931for MPLS,0.9950and0.9942for PLS,0.9947and0.9942for PCR,respectively.The differences of the results of the three methods for calibration were insignificant.The proposed method is rapid and accurate.It can be used to control the quality of milled rice.
出处
《分析测试学报》
CAS
CSCD
北大核心
2004年第4期43-45,共3页
Journal of Instrumental Analysis
基金
国家自然科学基金重大项目(39893350)
广东省自然科学基金资助项目(020384)
关键词
近红外透射光谱
定量分析
蛋白质含量
大米
Near infrared transmittance spectroscopy(NITS)
Quantitative analysis
Protein content
Rice
