Development Of Breath Detection System Based On Mid-infrared Laser Absorption Spectroscopy

The exhaled endogenous gas has the marker information which is closely related to the physiological metabolism of human body.By detecting the components and concentrations of the exhaled gas,the diseases can be diagnosed quickly and noninvasively.As a spectroscopic measurement way,tunable diode laser absorption spectroscopy?TDLAS?technique is suitable for real-time and on-line measurement of exhaled gas with high sensitivity,high selectivity and a simple and stable detection system.In this thesis,TDLAS was used to study the key technologies of breath detection based on mid-infrared?MIR?laser absorption spectroscopy.The main works are summarized as follows:Firstly,an exhaled gas detection system was designed and verified based on MIR laser spectroscopy.For the practical application of high sensitivity,fast response,convenient sampling and multi-component simultaneous detection in breath analysis,a 9.56?m quantum cascade laser?QCL?was selected as the light source,which can cover the absorption of ammonia,ethylene,carbon dioxide and other gases within the scanning range.And based on the research of modulation characteristics of the QCL,an integrated real-time online gas measurement system was developed including exhalation acquisition module based on the long optical path and small volume annular gas absorption cell.Then the first harmonic normalized the second harmonic wavelength modulation spectrum technology?WMS2f/1f?was used to realize the gas calibration free measurement.The feasibility and effectiveness of the system were verified by the measurement of ammonia and ethylene mixture.The results showed that the minimum detection limits of ammonia and ethylene were 7.4ppb and 44.7ppb respectively,and the corresponding optimal integration time was 4.52s and 3.12s respectively.Secondly,multi-component gas concentration inversion was performed based on multispectral fitting algorithm and BP neural network algorithm.The WMS-2f/1f simulation software was developed by using the spectral parameters of ammonia and ethylene in Hitran spectral database and the QCL characteristics measured in advance.Then the multi-component gas identification and concentration inversion were carried out for simulation data and actual experiment data by using multispectral fitting algorithm and the model of PCA feature extraction method combined with error back propagation artificial neural network?BPANN?classification algorithm respectively.It was proved that the two algorithms were feasible for qualitative and quantitative analysis of multi-component gas simultaneously.Through the comparison,the neural network algorithm has more advantages in the running time and concentration inversion accuracy of the algorithm.Thirdly,carbon dioxide in human breath was detected by the developed system.The CO₂with high content in exhalation was used as the target gas for the actual exhalation analysis and detection.The absorption spectral line with CO₂at1045.02cm^-1was selected as the target line,and then the real-time online measurement of CO₂concentration in the breath of the subjects was realized.Finally,the performance index of the respiratory CO₂detection system was analyzed.The results showed that the detection sensitivity of the system was 25.35ppm with the time resolution of 40ms,and it can be achieved 7.7ppm at an optimal integration time of 0.465s.Furthermore,the linearity of the detection system was determined to be 0.99988.In conclusion,the key technologies were studied of the multi-component concentration inversion and the breath measurement system design based on MIR laser absorption spectroscopy,which lays a foundation for breath analysis and disease diagnosis research.