A feature extraction method and a sampling system for fast recognition of flammable liquids with a portable E-nose

Abstract

A feature extraction method and a sampling system for fast recognition of flammable liquids with a portable E-nose were reported. The sampling system was mainly consisted of a sensor chamber and a micro pump. The sampling time of the sampling system taken in each sample was less than 1 s. The feature extraction method could be used to extract robust information for fast recognition with a calculating process. Four common flammable liquids and three incombustible drinks were well recognized with the portable E-nose by using the sampling system and the proposed feature extraction method. The recognition speed for each sample was 10 s. A common method which extracts the amplitude signal as features was adopted as a comparison with the proposed feature extraction method. Principal component analysis and discriminant function analysis were used in the data analysis. Successful classifications of 100% were gotten in the discrimination of flammable or incombustible species with both methods. In the original species classification, 85.7% of the samples with the proposed method were correctly classified, while only 57.1% of the samples with the comparing method were correctly discriminated. The results of data analysis well implied that the features extracted by the proposed method were more robust than the features extracted by the comparing method.