论文部分内容阅读
提出了一种应用同步荧光光谱技术无损快速鉴别料酒品牌的新方法。利用主成分分解法和小波变换法对料酒样品的同步荧光光谱信号进行了压缩处理,分别采用同步荧光光谱数据的第一主成分和小波细节系数为特征变量进行主成分分析和聚类分析,分类结果表明小波系数作为料酒的特征变量对料酒品牌分类正确率更高。利用偏最小二乘和径向基人工神经网络方法建立料酒品牌鉴别的定量分析模型,预测结果表明,两种判别模型对料酒品牌鉴别的准确率均达到100%。
A new method of nondestructive and rapid identification of cooking wine brand using synchronous fluorescence spectroscopy was proposed. The principal component analysis and wavelet transform were used to compress the synchronous fluorescence spectra of the samples. Principal component analysis and cluster analysis were performed using the first principal component and the wavelet detail coefficients of the synchronous fluorescence spectra as the characteristic variables. The results show that the wavelet coefficients as the characteristic variables of cooking wine have a higher correctness on wine brand classification. The partial least squares and radial basis artificial neural network method were used to establish the quantitative analysis model of the brand identification of cooking wine. The prediction results show that the accuracy of the two discriminating models is 100%.