期刊
MEASUREMENT
卷 199, 期 -, 页码 -出版社
ELSEVIER SCI LTD
DOI: 10.1016/j.measurement.2022.111458
关键词
Electronic nose; Combustible; harmful gases; Artificial intelligence; Machine learning; Feature selection
This paper introduces an intelligent E-Nose called GasCon-Enose to identify both the type and concentration level of combustible gases. GasCon-Enose extracts features from different domains and introduces a hybrid feature selection approach. The experimental results show that GasCon-Enose has high recognition accuracy and can be employed as a reliable platform for combustible gas sensing and concentration level identification in smart city environments.
Identifying combustible gases is vital to ensure a livable environment and prevent hazards affecting human safety. Most previous E-Nose studies aimed to either identify the gas or its concentration level. Moreover, they did not extract features from various domains or apply feature selection (FS) techniques. This paper introduces an intelligent E-Nose called GasCon-Enose to identify both the gas type and its concentration level using artificial intelligence. GasCon-Enose extracts features from time, frequency, and time-frequency domains from five metal oxide semiconductors (MOS) sensors. It fuses these features and investigates their impact on identification performance. Moreover, it introduces a hybrid FS approach. The results show the peak accuracies are 99.73% and 97.54% for gas type identification and concentration level identification after the hybrid FS. The performance of GasCon-Enose proves its reliability and ability to be employed as a platform for combustible gas sensing and concentration level identification in smart city environments.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据