改进神经网络在芝麻油掺伪检测中的应用

研究芝麻油掺伪检测问题,提高检测精度。由于成分复杂,掺伪后化学成分变化,直观难以检测。传统物理或化学芝麻油掺伪检测方法操作复杂,设备昂贵,存在不同程度缺陷。结合近红外光谱技术和神经网络优点,提出一种RBF神经网络-近红外光谱的芝麻油掺伪检测方法(NIR-RBF)。首先采用近红外光谱提取芝麻油样本的光谱信息,然后采用主成分分析提取光谱信息主要有效成分,最后将主要有效成分输入到神经网络进行学习,得到芝麻油掺伪检测结果。采用建立的模型对掺入不同类型植物油的芝麻油进行检测,结果表明,相对于其它芝麻油掺伪检测方法,NIR-RBF提高了检测精度和速度,降低了检测误差,是一种快速、有效的芝麻油掺伪检测方法。

Application of Improved Neural Network in Sesame Oil Adulteration Detection

Study sesame oil adulteration detection problems to improve the accuracy of detection.The traditional physical or chemical sesame oil adulteration detection methods need complex operation,expensive equipment.Combined the advantages of near infrared spectroscopy technology and neural network,the paper proposed a RBF neural network-near infrared spectroscopy sesame oil adulteration detection method(NIR-RBF).First,sesame oil sample spectral information was extracted by near infrared spectroscopy.Then,the principal component analysis was used to extract main effective components of spectral information Finally,the main effective components were input to the neural network for learning,and the detection results of sesame oil adulteration were acquired.Using the established model to detect the sesame oil incorporating different types of plant oil,the results show that,compared with other sesame oil adulteration detection method,NIR-RBF improves the detection precision and speed,reduces the error of detection,and is a fast and effective method for detection of sesame oil adulteration.