Study On Quantitative Analysis Of Water Content Of Ammonium Perchlorate Based On Terahertz Spectrum

Ammonium perchlorate(AP)faces the demand for water content detection in production,storage,transportation and use.The current method of AP moisture content detection has the defect of long detection period.Terahertz time domain spectroscopy has the characteristics of fast,nondestructive,safe and high signal-to-noise ratio,and water has strong absorption to Terahertz.The researches showed that terahertz time domain spectroscopy has great potential in the field of water content detection.In this thesis,the moisture content of AP was studied based on terahertz time-domain spectroscopy.The main contents are as follows:Firstly,the thesis researched the preparation method and water content calibration method of water-bearing AP samples and the influence of water distribution uniformity,sample temperature,sample box surface roughness,the position of sample location influence on THz spectrum.The influence of reflection peak on test results was studied.The THz spectrum of the water-bearing AP samples was measured by the transmission terahertz time-domain spectrum system,and the absorption coefficient and refractive index of the water-bearing AP samples were extracted.Secondly,PCR,PLSR and SVR analysis methods were used to achieve quantitative analysis of AP moisture content,and the SVR regression analysis method based on absorption coefficient has the best prediction effect,the R~2and RMSE were 0.8528,0.0014,respectively.According to the Beer-Lambert Law and Effective medium theory,a prediction model of AP water content based on absorption coefficient was constructed.According to Effective medium theory and equal optical path,a prediction model of AP water content based on refractive index was constructed.The two prediction models need to weigh the sample.In addition,a new prediction model of AP water content based on absorption coefficient and refractive index was proposed,which does not need to weigh the sample under the condition of ensuring the measurement accuracy.Finally,based on the prediction models of AP moisture content,an optical parameter extraction method based on the theoretical model was proposed;According to the intercept and slope of linear fitting between absorption coefficient and refractive index of water-containing AP samples and water content,an optical parameter extraction method based on data fitting is proposed.The results showed:the optical parameter extraction method based on data fitting is superior to the optical parameter extraction method based on theoretical model,the water content prediction model based on absorption coefficient has the best effect,and its R~2and RMSE were 0.9609 and 7.30E-04,respectively.The research showed that the Terahertz time domain spectroscopy has great potential in the field of low water content and high precision detection,and can provide a rapid,nondestructive,safe and high precision detection method and for the detection of moisture content in AP and similar solid powders.