基于主成分分析和神经网络的近红外光谱烤烟产地判别

Discrimination of Flue-cured Tobacco Producing Area by Using Near Infrared Spectroscopy(NIR)Based on Principal Component Analysis and Artificial Neural Network Model

为快速准确地对烤烟样品进行产地判别,用近红外光谱对湖南、福建、广东、云南和贵州5个省2015年的459个样品进行扫描,经光谱预处理后,用主成分分析法对烤烟产地进行分析并建立BP人工神经网络模型。结果表明:该模型对未知的92个样品的预测准确率达98.91%。

459flue cured tobacco samples from Hunan,Fujian,Guandong,Yunnan and Guizhou were scanned by near infrared spectroscopy(NIR)in2015and the BP artificial neural network model was established based on principal component analysis of producing areas after spectral pretreatment to rapidly and accurately discriminate producing area of flue cured tobacco samples.Results:The predicting accuracy reaches98.91%when the established BP artificial neural network model is used to discriminate producing area of unknown92samples.