Research On Gas Recognition And Concentration Detection Algorithm Based On Infrared Spectrum

In recent years,with the development of semiconductor lasers and the continuous deepening of laser spectroscopy,research on measurement methods based on infrared semiconductor laser absorption spectra has attracted great attention from researchers in various fields in various countries.Infrared spectroscopy has many advantages such as rapid,high-efficiency,non-destructive and suitable for on-line analysis.Therefore,its development prospect is broad.With the rapid development of aerospace technology,gas detectors are one of the important sensors that ensure the safety of astronauts and the stable operation of space stations.Gas sensors based on infrared spectroscopy have become one of the important components of aerospace modules due to their miniaturization and integration.Infrared gas sensors are mainly used in the monitoring of changes in harmful gas components in passenger compartments,fuel combustion during the operation of spacecraft engines,the determination of gas or liquid leakage in key parts of spacecraft,and detection of planetary atmospheres during deep space exploration.Product quality,safety,detection of the surrounding environment,and astronaut's physical health provide very important reference data.Therefore,a number of foreign aerospace research institutes are strongly supporting the research and development of various types of gas sensors suitable for work on spacecraft,and have been applied in the aerospace work environment for many times.In order to meet the above requirements,various types of sensors have been studied in-depth abroad.In order to more quickly and accurately identify the gas components and their corresponding concentrations in the environment.Based on the characteristics of the gas spectrum and the distribution of infrared absorption spectra,the principle of the detection of the spectrum was studied.The time-domain analysis method was used to obtain the five gases infrared band absorption characteristics,including NH₃?HCl?CO?NO?CH₄.A concentration recognition algorithm based on multivariate interpolation radial basis function method?RBF neural network?was proposed.It was processed by Matlab software.The model prediction and model setting were used to obtain relevant prediction models.Further,in order to illustrate the comparison of RBF neural network recognition effect,this paper also studied the BP neural network algorithm,the same use of sample data to predict function model training,and get the corresponding prediction function.Predictive model data is predicted by the prediction model,and the five gas concentrations are well predicted.By calculating the deviation between the predicted value and the true value,the generalization ability of the network can be evaluated.Finally,the performance evaluation parameters such as relative error and coefficient of determination all show that the RBF neural network algorithm can be better applied in gas concentration recognition.