基于电子鼻技术的葡萄酒分类检测方法

Classification of Wine by Electronic Nose

为了简单准确的检测葡萄酒的种类,建立了电子鼻检测系统。以三种具有相似气味的葡萄酒的种类识别为实验背景,根据葡萄酒散发的气味合理的选用了八个气敏传感器。利用主成份分析方法对传感器阵列进行优化,最后确定选用四个传感器为最终的传感器阵列,并借助Fisher判别分析方法检验其效果。使用SVM算法及BP算法分别对不同训练样本数的葡萄酒做对比实验。实验结果表明,基于PCA-SVM模式识别算法有很高的识别精度,很强的分类能力,而且在小样本分类识别实验中有着潜在的优势。

An electronic nose system was developed for detecting different class of wines. Identification experi- ments for three kinds of wine with similar odor were carried out. According to the aroma of wine, the electronic nose system has been developed with an array of four gas sensors, which was optimized from initial eight by PCA method. And then, the effect of the optimization for three wines was discriminated by Fisher discriminant analysis. Finally, there were comparative emulation experiment for different number of training samples of wine by using SVM algorithm and BP network algorithms. The experimental results show that PCA-SVM-based pattern recognition algo- rithms has high recognition accuracy, strong classification capability, and there are potential advantages in small sample classification and recognition experiments.