Study Of Stabilizer Measurement Method Based On Diffuse Reflectance Spectroscopy

Evaluation of weapon and equipment health status is necessary.Traditional destructive evaluation methods are costly,thus non-destructive evaluation methods are valued,where verification of sufficient concentration of stabilizer in the propellant can be used as an evaluation indicator.Common stabilizer detection methods have problems such as time-consuming analysis and frequent sampling.Therefore,the combination of UV-Vis diffuse reflectance spectroscopy and fiber optic sensing for qualitative and quantitative study of stabilizer and thus propellant state assessment has the advantages of non-destructive,rapid and in-situ measurement,and has promising applications.In this paper,three types of stabilizers,2-Nitrodiphenylamine(2-NDPA),N-Methyl-4-nitroaniline(MNA)and N-Ethyl-4-nitroaniline,were studied,focusing on the measurement technique of stabilizers based on diffuse reflectance spectroscopy.Design and build a diffuse reflection spectral detection system based on fiber optic sensing.A qualitative and quantitative study of stabilizers was performed by diffuse reflectance spectroscopy of samples combined with chemometric methods to verify the feasibility of the technical approach.The relevant studies and the achievements are as follows:1.Firstly,the theory involved in diffuse reflectance spectroscopy is derived to provide theoretical support for experimental studies.Designed and built a spectral detection system based on the principle of diffuse reflection spectral detection,the system mainly includes light source module,spectral detection module and optical transmission module,etc.The effects of sample thickness,clamping position and pressure on the spectra were investigated in depth.The results show that the thickness of the sample to be measured has a large influence on the spectral measurement,and the sample thickness should be kept greater than 3 mm in the actual measurement.2.The support vector machine(SVM)classification recognition principle and programming implementation method are studied in depth for the diffuse reflection spectral characteristics of three types of stabilizers.A classification model was developed using SVM to classify three types of tranquilizer samples with concentrations between 0.1%and 1%based on diffuse reflectance spectra.The results show that the classification algorithm can be applied to identify the types of stabilizers contained in the samples according to the diffuse reflectance spectrum,and the recognition accuracy of the test set can reach 97.4%.3.Experimental studies on the detection of diffuse reflectance spectra of three types of stabilizers were carried out separately.The corresponding characteristic spectral intervals,2-NDPA(500-560 nm),MNA(460-530 nm),and N-Ethyl-4-nitroaniline(460-490 nm),were determined.The principles of partial least squares(PLS)and support vector regression(SVR)were explored to develop corresponding quantitative models for the concentrations of the three stabilizers.The results showed that based on the characteristic band(characteristic spectral interval)modeling,the SVR model had the best prediction for 2-NDPA and N-Ethyl-4-nitroaniline with R~2 of 0.9844 and 0.9976 and RMSE of 0.000272 and 0.000199,respectively.The PLS model has the best prediction for MNA with R~2 and RMSE of 0.9936 and 0.000310,respectively.Based on the characteristic wavelength modeling,2-NDPA(532 nm and 560 nm),MNA(470 nm and 532 nm),and N-Ethyl-4-nitroaniline(470 nm),the SVR model had the best predictions with R~2 of 0.9966,0.9948,and 0.9959,respectively,and RMSE of 0.000256,0.000254,and 0.000271.The above results provide a reliable reference for rapid and accurate quantitative detection of stabilizer concentration.