【摘要】 以来自不同地点的126份薯块根样品为材料,应用近红外光谱技术(NIRS),建立甘薯黄酮含量近红外反射光谱分析数学模型,并对模型预测结果的准确性进行了评价。结果显示,甘薯块根黄酮含量的近红外光谱模型校正决定系数(RSQ)为0.903,校正标准误差(SEC)为0.172。交叉验证决定系数(RSQV)为0.851,交叉验证标准误差(SECV)为0.214。模型对样品NIRS的预测值与其相应的化学值有较好的相关性,此模型可用来预测甘薯黄酮含量,在甘薯优质育种和品质分析中具有应用价值。更多还原

【Abstract】 Predicted mathematic model for analysis of flavones in 126 samples of sweetpotato storage roots from different locals was established by near infrared reflectance spectroscopy(NIRS). The veracity of model was evaluated by the determination coefficients of validation(RSQ) and the standard errors of calibration(SEC),the correlation coefficients of internal cross validation test (RSQV) and the corrected standard errors of performance(SECV),which was 0.903,0.172,0.851 and 0.214,respectively,the better relativities of the data for flavones from NIRS model and chemical measurement were concluded.The data of quantitative analysis for flavanone from NIRS model can be accepted.The model can be used in attributional analysis and quality breeding of sweetpotato.更多还原