关键词: 水稻, 蛋白含量, 近红外光谱技术, 校正模型

Abstract:

Six predicted mathematic models for analysis of protein content in brown and milled rice were established, with the technique of near infrared reflectance spectroscopy (NIRS) and partial least square (PLS) algorithm. The protein content of brown rice and milled rice was determined by chemical methods. The different predicted models were established with the near infrared spectroscopy of paddy, whole grain and rice flour, which was to research the effect of sample forms on predicting veracity of NIR models. The determination coefficients (R_C²) of calibration of NIR models of brown rice protein content for paddy, whole grain and flour were 0.893, 0.971 and 0.987 respectively, the root mean square errors of calibration (RMSEC) was 0.507, 0.259 and 0.183. The RC2 of NIR models of milled rice protein content for paddy, whole grain and flour were 0.897, 0.984 and 0.986，the RMSEC was 0.497, 0.186 and 0.190 respectively. The veracity of models was estimated by the determination coefficients (R_CV²) and the root mean square errors (RMSECV) of cross-validation, the determination coefficients (R_V²) and the root mean square errors (RMSEV) of external validation. The R_CV² of NIR models of brown rice protein content were 0.865, 0.962 and 0.984 respectively，the RMSECV 0.557, 0.290 and 0.205. The R_CV² of NIR models of milled rice protein content were 0.845, 0.951 and 0.979，the RMSECV 0.594, 0.316 and 0.233. The R_V² of the NIR models for brown rice protein content were 0.683, 0.801 and 0.939，the RMSEV 0.962, 0.799 and 0.434. The R_V² of the NIR models for milled rice protein content were 0.673, 0.921 and 0.959，the RMSEV 0.976, 0.513 and 0.344. From the above results, we concluded that the veracity of flour NIR models is the best, that of whole grain better, while that of paddy poor. The NIR method can substitute the chemical method and be widely used in the breeding and quality analysis of rice.

Key words: Rice, Protein content, Near infrared reflectance spectroscopy technique, Calibration model