Research And Design Of The Breath Detector Based On The Sensor Array

The early clinical symptoms of most diseases are not obvious,which makes many patients miss the best opportunity for diagnosis and treatment.Human exhalation contains the products of metabolism,which can reflect the health status of human body to a certain extent.Through the detection of certain disease-related gas markers in the breath,it is expected to realize the initial screening and early diagnosis of possible diseases in human body.Compared with the traditional gas chromatography,the breath detection method based on array sensor has the advantages of low cost,simple operation,and small volume,which is very suitable for the development of portable gas detection instrument.This paper studies and designs a breath detection instrument based on array gas sensor,which is composed of sensor array,sensor interface circuit and pattern recognition system(algorithm).It can recognize acetone,ethanol and other marker gases related to lung cancer.According to the design requirements of gas identification and sensor selection specifications in this paper,four metal oxide(MOS)gas sensors are finally selected to form a 2*2 sensor array.The overall sensitivity of the traditional interface circuit based on the resistance series voltage division method is affected by the load resistance,the linearity is poor,and the measurement accuracy and dynamic range of the gas-sensitive resistance are limited.It is difficult to match different MOS gas sensors in the front-end sensor array.To solve this problem,the paper studied and designed a multi-range automatic detection circuit based on constant current source,and realized automatic selection of interface circuit range through STM32 master chip.And in the resistance range of 100Ω-200KΩ,the test results show that the circuit has good linearity in the whole measurement range and the overall error is less than 3%.The current gas concentration data of the instrument is displayed by the serial touch screen(simple human-computer interaction can be realized through touch),and transmitted to the upper computer by the serial communication module for subsequent pattern recognition.The system software program is developed based on Keil software and written in C language.The main tasks are AD multi-channel signal acquisition,automatic selection of interface circuit range,touch screen display and so on.Finally,the mixed gas data samples collected by the system are tested.The pattern recognition algorithm used is studied.Kernel Principal Component Analysis(KPCA)was used to extract features from data,and Support Vector Machine(SVM)classifier was used for qualitative identification of gas samples.The BP neural network Algorithm optimized by Genetic Algorithm(GA)is used to achieve quantitative detection of gas concentration.Acetone and ethanol were used as the target gas in the laboratory,and the corresponding small concentration gas samples were configured to detect the mixed gas in the designed instrument.The experimental results show that the average recognition rate of binary gas mixture is 97.72%,and the average error of quantitative analysis of mixed gas is only 2.28%.After testing,the instrument has realized the function of collecting,detecting,and analyzing binary mixture gas of ethanol and acetone,which provides a foundation for the early detection of subsequent diseases.