School of Chemistry, Sun Yat-Sen University, Guangzhou 510275, China.
Technology Center, China Tobacco Guangdong Industrial Co., Ltd., Guangzhou 510385, China.
Sensors (Basel). 2018 Aug 13;18(8):2658. doi: 10.3390/s18082658.
The aim of this study is to improve the discrimination performance of electronic noses by introducing a new method for measuring the similarity of the signals obtained from the electronic nose. We constructed abstract odor factor maps (AOFMs) as the characteristic maps of odor samples by decomposition of three-way signal data array of an electronic nose. A similarity measure for two-way data was introduced to evaluate the similarities and differences of AOFMs from different samples. The method was assessed by three types of pipe and powder tobacco samples. Comparisons were made with other techniques based on PCA, SIMCA, PARAFAC and PARAFAC2. The results showed that our method had significant advantages in discriminating odor samples with similar flavors or with high VOCs release.
本研究旨在通过引入一种新的方法来测量电子鼻信号的相似性，从而提高电子鼻的区分性能。我们通过对电子鼻的三向信号数据阵列进行分解，构建了抽象气味因子图谱（AOFM）作为气味样本的特征图谱。引入了一种用于衡量两向数据相似性的方法，以评估来自不同样本的 AOFM 的相似性和差异性。该方法通过三种类型的烟管和烟粉样品进行了评估。并与基于 PCA、SIMCA、PARAFAC 和 PARAFAC2 的其他技术进行了比较。结果表明，我们的方法在区分具有相似风味或具有高 VOCs 释放的气味样品方面具有显著优势。
