NIRS定量分析油菜种子含油量、蛋白质含量数学模型的创建

Establishment of Math Models of NIRS Analysis for Oil and Protein Contents in Seed of Brassica napus

用近红外光谱 (NIRS)分析油菜品质。采用残余法测试了近 2 116份甘蓝型油菜品种资源种子的含油量 ,用近红外仪采集数据 ,选择 12 88份代表性样品 ,建立了数学模型。用该模型测试 96份待测样品 ,其NIRS的测试值与残余法测试的油菜种子含油量实测值相关系数为 0 .950 3 ,相对误差小于 3 .5% ,用凯氏定氮法测试了 63 7份油菜籽饼粕的蛋白质含量 ,选择 168份代表性样品 ,建立数学模型。 3 0份样品检测模型 ,NIRS测试值与凯氏定氮法测试的油菜籽饼粕蛋白质含量的实测值相关系数为 0 .9515,相对误差小于 6%。结果表明 ,这 2个数学模型已经可用来准确、快速、无污染。

In this study, 2 116 seeds of B.napus were analyz ed on oil content by the residual method. Perten near-infrared reflectance spectr oscopy(NIRS) used 1 288 seeds to establish math model of oil content. 96 sampl es were used to test this model, NIRs predictions compared to reference results agree excellently, the relative coefficient (R) was 0.9503 , relative mean de viation was below 3.5%. 637 rapeseed meal were determined on protein content by Mico-Kjeldahl method. Perten NIRS used 168 to establish math model of protein content. 30 samples were used to test, NIRS predictions compared to reference re sults agree excellently, the relative coefficient (R) was 0.9515, relative mean deviation below 6%. So NIR technique can be used to determine accurately and fle etly oil and protein content of B.napus without free pollution and depletion in short time.