期刊:Proceedings of IEEE Sensors [Institute of Electrical and Electronics Engineers] 日期:2024-10-20卷期号:: 1-4被引量:2
标识
DOI:10.1109/sensors60989.2024.10784840
摘要
In this article, a fully integrated Electronic-nose (E-nose) system for gas recognition is proposed. Integrated with four Micro-Electro-Mechanical systems (MEMS) metal-oxide (MOX) gas sensors and a programmable interface circuit, the E-nose system can measure, process, analyze and output the gas types without external devices. In order to further improve the gas-sensitive feature datasets of different odors, the E-nose microsystem can also virtually expand up to 256 half-virtual gas-sensitive pixels by operating temperature modulation and measuring voltage regulation. Similar to current image sensor, this E-nose microsystem can continuously generate a gas grayscale image with 16$\times 16$gas-sensitive pixels at a frequency of 0.0625 Hz. With a lightweight convolutional neural network (LCNN) algorithm embedded in the E-nose system, it can directly and precisely identify nine gas types based on gas-sensitive grayscale image.