摘要
利用傅里叶近红外光谱技术建立水蜜桃货架期糖度无损检测方法。采用多元散射校正、标准正态变量转换、Savitzky-Golay一阶导数和卷积平滑对原始光谱进行预处理,利用偏最小二乘法建立模型,并根据相关系数选取合适的建模波段。结果表明:多元散射校正、Savitzky-Golay一阶导数和卷积平滑结合处理能够有效地消除光谱基线平移和偏移等现象,提高信噪比,构建的模型预测效果较好。为了提高运算速度,设置相关系数绝对值阈值为0.55时,建模的波长数为389,构建的模型预测效果也较好,校正组模型和预测组模型的决定系数分别为0.860和0.853,内部交叉验证均方差(RMSECV)和预测均方根偏差(RMSEP)分别为0.651和0.732°Brix。同时比较了PLS-NIR、PCA-NIR、PLS-BPANN-NIR、PCA-BPANN-NIR和PCA-SVM-NIR模型对糖度的预测性能,结果表明PLS-NIR模型对货架期水蜜桃糖度预测效果最好。
In this paper,the Fourier near infrared spectroscopy technology was used to explore nondestructive testing methods to detect the sugar content of juicy peach during shelf life.Four pre-processing methods i.e.mutiplicative scatter correction,transformation of standard normal variate,Savitzky-Golay first derivative and Savitzky-Golay smoothing were used and a model was set by the partial least square(PLS)method.On this basis,the correlations between the values of original spectrum after the pretreatment and the sugar content of peach were analyzed,based on which the suitable modeling bands were selected.The results showed that multiplicative scatter correction,Savitzky-Golay first derivative and Savitzky-Golay smoothing combination treatment could effectively eliminate the phenomenon such as spectral baseline shift and migration,and it could improve signal-to-noise ratio and construct the best predicted model.In order to raise the operating speed,the absolute value threshold of correlation coefficient was 0.55 and the number of wavelengths was 389.Then the model also showed better predicted results,the correlation coefficient(r)of calibration and validation model were 0.860 and 0.853 respectively,and the root mean square error of cross validation(RMSECV)and the root mean square error of prediction(RMSEP)were 0.651 °Brix and 0.732 °Brix respectively.In addition,the models based on PLS-NIR,PCA-NIR,PLS-BPANN-NIR,PCA-BPANN-NIR and PCA-SVM-NIR were compared with each other,and the model based on PLS-NIR was the best.
出处
《南京农业大学学报》
CAS
CSCD
北大核心
2013年第4期116-120,共5页
Journal of Nanjing Agricultural University
基金
国家自然科学基金项目(31101282)
中央高校基本科研业务费专项资金项目(KYZ201120)
国家公益性行业(农业)科研专项(201303088)
江苏省高校优势学科建设工程项目
中农-南农青年教师开放基金项目(NC2008004)
关键词
近红外光谱
水蜜桃
糖度
货架期
检测
near infrared spectroscopy
juicy peaches
sugar content
shelf life
detection
