红色和黑色种皮花生籽仁含油量检测模型的构建  

Construction of Near-infrared Spectra Models for Oil Content Prediction in Red and Black Testa Peanut

作　　者：刘雨 侯名语[1] 崔顺立[1] 刘盈茹 李秀坤 刘怡诺 刘立峰[1] LIU Yu;HOU Ming-yu;CUI Shun-li;LIU Ying-ru;LI Xiu-kun;LIU Yi-nuo;LIU Li-feng(College of Agronomy,Hebei Agricultural University/State Key Laboratory of North Chinafor Crop Improvement and Regulation,Baoding 071001,China)

机构地区：[1]河北农业大学农学院/省部共建华北作物改良与调控国家重点实验室,河北保定071001

出　　处：《花生学报》2025年第1期79-86,116,共9页Journal of Peanut Science

基　　金：国家现代农业产业技术体系建设专项(CARS-13);河北省现代农业产业技术体系建设专项(HBCT2024040205);河北省现代种业科技创新专项(21326316D-2);河北省农业成果转化项目(V1648016174968)。

摘　　要：含油量是花生品种的重要品质指标,高效无损检测花生含油量是花生种质鉴定及品种选育的重要研究内容。粉色种皮花生含油量近红外检测模型已得到广泛应用,而红色和黑色种皮花生含油量的近红外模型的构建及育种应用较少。本研究选用98份黑色种皮花生和92份红色种皮花生为材料构建特色花生含油量近红外检测模型。98份黑色种皮花生含油量为40.05%~53.6%,92份红色种皮花生含油量为36.09%~51.37%。黑色种皮花生光谱值预处理方法为平滑滤波导数(Savitzky-Golay Derivative)、标准正态变量变换(SNV)及去趋势(De-trending)的组合,红色种皮花生光谱值预处理方法为平滑滤波导数(Savitzky-Golay Derivative)与去趋势(De-trending)的组合。采用最小二乘偏回归检验方法构建定标模型,黑色种皮花生定标模型的决定系数(R^(2))为0.9191,RMSEC为0.712,外部验证决定系数(R^(2))为0.93;用粉色种皮花生、红色种皮花生含油量近红外模型进行交叉验证,决定系数分别为0.2335、0.0156。红色种皮花生定标模型的决定系数(R^(2))为0.839,RMSEC为1.437,外部验证决定系数(R^(2))为0.805;用粉色种皮花生、黑色种皮花生含油量近红外模型进行交叉验证,决定系数分别为0.241、0.079。验证结果表明红色、黑色花生含油量的近红外检测模型准确可靠,可用于特色种皮花生含油量的品种选育。Oil content serves as a crucial quality index for peanut varieties. Efficient non-destructive testing of peanut oil content is important for identifying and selecting peanut germplasms with high oil content. The near-infrared detection model for oil content in peanuts with pink testa has been widely used, while the construction and application of near-infrared models for oil content in peanuts with red and black testa are relatively limited. In this study, 98 samples of black peanuts and 92 samples of red peanuts were selected as materials to establish near-infrared detection models for oil content. The oil content of the 98 black peanuts ranged from 40.05% to 53.6%, and the oil content of the 92 red peanuts ranged from 36.09% to 51.37%. A combination approach of Savitzky-Golay Derivative, Standard Normal Variable(SNV) and De-trending were used by preprocessing the spectral values of black peanuts. For red peanuts, the preprocessing method was comprised of Savitzky-Golay Derivative and De-trending. The calibration models were constructed utilizing the least squares partial regression method. The black peanut calibration model achieved a coefficient of determination(R~2) of 0.9191, with an RMSEC of 0.712 and an external validation(R~2) of 0.93. Cross-validation of oil content in black peanuts using near-infrared models for pink and red peanuts revealed a coefficient of determination of 0.2335 and 0.0156 respectively. The red peanut calibration model yielded an R~2 of 0.839, an RMSEC of 1.437, and an external validation R~2 of 0.805. Additionally, cross-validation of oil content in red peanuts with near-infrared models for pink and black peanuts showed coefficient of determination 0.241 and 0.079 respectively. These results showed that the testa color-specific near-infrared detection models for oil content were accurate and reliable, which can be used for the identification of oil content in peanut breeding with red and black testa color.

关 键 词：黑色种皮花生 红色种皮花生 含油量 近红外模型 

分 类 号：S565.2[农业科学—作物学] Q547[生物学—生物化学]