E-nose System using CNN and Abstract Odor Map to Classify Meat Freshness

In this paper, a CNN-based electronic nose system using time series features is proposed for classifying meat freshness. The electronic nose consists of MEMS gas sensor array, data acquisition device, and a CNN-based classification unit. The gas sensors manufactured by MEMS own the features of low power consumption and small size. The detection acquisition device is realized in the form of discrete components on a printed circuit board (PCB). Time series feature extraction is introduced to extract transient information for data features extension. Odor transient and steady-state information are combined to construct abstract odor map. Abstract odor map is sent to CNN-based classification unit to classify meat. Finally, we demonstrate that time series feature extraction can improve the meat classification accuracy by 6.3% and achieve 98.4% of meat freshness accuracy.