Optimal gas sensor combination selection method for low cost machine olfaction applicated in food discrimination

In this paper an electronic nose system is proposed, together with odor-related gas information data processing and recognizing methods and an optimal gas sensor combination selecting procedure. The constructed electronic nose system consists of three parts, including a gas sensor chamber, data collecting circuits loaded with gas sensors and odor-related information data processing AI-driving programs. Utilizing the constructed electronic nose system, we accomplish the classifying missions of most 7 fruits and 8 vegetables to classifying accuracy of 96.7% using SVM classification algorithm and 10 sensor units. All probable gas sensor combinations are checked by the standard of their classifying performance to find the gas sensor combination match best when working together. The outcome turned to find that a specified gas sensor combination composed of 4 sensor units work best in the fruits and vegetables classification. This outcome suggests that these 4 sensor units are most suitable utilizing in related fruits and vegetables classification mission.