基于多域特征融合的手持式电子鼻白酒识别

Multi-Domain Feature Fusion Based Chinese Liquor Recognition Using a Hand-Held Electronic Nose

为提升手持式电子鼻辨别白酒的准确率,本文提出了一种基于多域特征融合的识别方法。首先,分别基于统计学分析、小波包分析和一对多共同空间模式(One Versus Rest Common Spatial Pattern,OVR-CSP)提取信号的时域、时频域和空域特征。其次,基于特征加权的方法将时域、时频域和空域特征进行融合。利用自制的手持式电子鼻采集了6种白酒样本并进行识别实验,结果显示,采用支持向量机(Support Vector Machine,SVM)作为分类器时,与单独使用时域、时频域和空域特征相比,所提多域融合特征方法的识别准确率分别提高了9.83%、8%和1.5%。进一步比较了SVM、K近邻(K-Nearest Neighbour,KNN)和BP神经网络(Back Propagation Artificial Neural Network,BP-ANN)三种分类识别方法识别准确率和运行时间,结果表明,KNN算法的用时最短,且识别率较高。

In order to improve the accuracy of the hand-held electronic nose(e-nose)for distinguishing Chinese liquor,a recognition method based on multi-domain feature fusion is proposed.First,the time domain,time-frequency domain and spatial domain features of the signals are extracted based on statistical analysis,wavelet packet analysis and one versus rest common spatial pattern(OVR-CSP),respectively.Second,the time domain,time-frequency domain and spatial domain features are fused by a weighting method.A self-made hand-held e-nose was used to collect 6 kinds of Chinese liquor samples and perform recognition experiments.Taking the support vector machine(SVM)as the classifier,and the results show that the recognition accuracy of the proposed multi-domain feature fusion method has been increased by 9.83%,8%and 1.5%,compared with time domain,time-frequency domain and spatial domain features alone,respectively.The recognition accuracy and running time of the system based on SVM,K-nearest neighbour(KNN),and back propagation artificial neural network(BP-ANN)were compared,and the results show that the KNN algorithm takes the shortest time and has a relatively high recognition accuracy.