摘要
采用超声法测量牛奶成分,并对牛奶超声特性进行了探讨,分别通过偏最小二乘法、非线性偏最小二乘法和非线性主成分回归,建立脂肪和蛋白质的校正模型。结果表明,用非线性偏最小二乘法建模预测的结果最好,脂肪的预测标准偏差RMSEP值为0.2533,蛋白质的预测标准偏差RMSEP值为0.2303,而与标准化学法相比,其预测结果不存在明显差异。
In order to validate the feasibility of measuring the composition contents using ultrasonic method, the ultrasonic characteristics of milk were investigated in detail. Then, based on ultrasonic velocity and ultrasonic attenuation, the calibration models for fat and protein were constructed through partial least squares (PLS), nonlinear partial least squares (NLPLS) and nonlinear principal component regression (NLPCR) , respectively. The experimental results showed that the best root mean square error of prediction (RMSEP) for fat and protein were 0.253 3 and 0.230 3, which were both obtained by NLPLS method. Furthermore, there was also no obvious difference between the results measured by standard chemical method and ultrasonic method through t-testing, which indicted that the contents of fat and protein in milk can be obtained using ultrasonic method.
出处
《农产品加工(下)》
2009年第9期20-22,共3页
Farm Products Processing
基金
河南工业大学博士基金项目(2009年)
关键词
脂肪
蛋白质
超声检测
衰减
声速
fat
protein
ultrasonic measurement
attenuation
ultrasonic velocity