Research On Signal Processing Algorithm Of Infrared Gas Sensor

Gases are widely present in various production and living environment.The flammable,explosive,toxic and harmful properties of some gases not only bring potential dangers to personal safety,but also affect the atmospheric environment.Therefore,rapid and accurate detection of gas concentration,inversion of gas parameters,and location of gas leakage sources can not only effectively avoid the occurrence of safety accidents,but also monitor the concentration of harmful gases in the atmosphere in real time.Among many gas detection methods,infrared absorption spectroscopy method is widely used due to its good selectivity,high sensitivity,short response time and long life.In order to suppress the noise in the system or environment,improve the inversion accuracy,and effectively locate the leak source,it is necessary to integrate various signal processing algorithms with infrared absorption spectroscopy technology to highlight its fast response and high accuracy.In this paper,a modern adaptive filtering algorithm is used to suppress noise with unknown statistical characteristics in infrared sensor signals;a neural network algorithm is used to invert gas parameters;a weighted centroid and differential evolution algorithm are used to locate gas leakage source.First,the gas detection algorithm based on adaptive filtering technology is studied.The absorption spectrum of methane molecules in the mid-infrared band was selected.The optical part of the system structure consists of two broadband thermal light sources,two semi-ellipsoid condensers,two dual-channel pyroelectric detectors and an air chamber.The electrical part consists of a constant current drive circuit module,a data acquisition(DAQ)card and a Lab VIEW data processing platform.A gas detection algorithm model based on adaptive filtering technique is established,and the algorithm is improved based on the numerical simulation results.Methane experiments were carried out.For standard gas samples in the concentration range of 0-10 k ppm,the relative error reached a maximum value of 13.09%.The detection accuracy was 165 ppm when the integration time was 3 s,and the detection accuracy was reduced to 2 ppm when the integration time was increased to 1050 s.At 0 ppm,the stability of the system with noise of different frequencies is measured for 360 min,it is verified that the concentration value using the adaptive filtering algorithm is closer to the true value,and the concentration standard deviation after filtering is smaller with the noise frequency increasing.Then,the gas parameter inversion algorithm based on neural network is studied.The absorption spectrum of methane molecules in the near infrared band is selected.The optical part of the sensor system is composed of a near infrared distributed feedback(DFB)laser,a gas chamber,and an Indium Gallium Arsenide(IGA)detector,the electrical part is composed of a laser current driver,a data acquisition card and a Lab VIEW data processing platform,and the gas pre-preparation part is composed of a mass flow controller,a pressure controller and a vacuum pump.A wavelet transform method is used to suppress system noise.A gas parameter inversion algorithm model based on neural network is established,the network structure,transfer function and learning function are optimized,and the algorithm is numerically simulated on MATLAB.A methane experiment was carried out,and the stability of a standard methane gas with a concentration level of 5000 ppm was tested for 30 min.The measured mean value of 12CH4 isotope gas concentration was 4957.00 ppm,and the standard deviation was 6.76 ppm;The measured mean value of 13CH4 isotope gas concentration is 55.80 ppm,and the standard deviation is 0.05 ppm;The measured mean value of the chamber pressure is 103.51 Torr and the standard deviation is 1.73 Torr;The measured mean value of 13CH4 isotope abundance is 2.13‰,and the standard deviation is 0.81‰.Finally,the algorithm for locating the source of gas leakage is studied.The gas turbulence diffusion model is selected,and the gas diffusion equations in windless and windy environments are given respectively.Two algorithms,weighted centroid and differential evolution,are used as leak source location algorithms.A human-computer interaction interface based on MATLAB is designed.By setting the gas leakage source parameters,environmental parameters and monitoring node coordinates,the concentration distribution map,gas diffusion map,positioning results,etc.can be visually displayed.Carrying out the positioning simulation,the results show that: before the gas diffusion balance,the time required to reach the minimum positioning error using the differential evolution algorithm is significantly shorter than the time required to use the weighted centroid algorithm;after the gas diffusion equilibrium,the positioning error is affected by many factors such as the distribution of monitoring nodes,the concentration value at the monitoring nodes,and the positioning algorithm used.If the more compact the distribution of monitoring nodes,the more accurate the concentration value obtained at the nodes,and the more intelligent the positioning algorithm,the smaller the positioning error is.The main innovations of this paper are as follows:The gas detection algorithm based on adaptive filter technology is designed with four channel structure.The algorithm is improved according to the numerical simulation results,and the methane gas detection experiment is carried out to verify the effect of the algorithm.