Research On Terahertz Spectrum Data Processing And Quantitative Analysis

Terahertz time-domain spectral technology is more and more widely used in material testing. This article in view of the terahertz time-domain spectrum data processing and quantitative regression problems, investigated the batch calculation for terahertz optical parameters, reflection peak removal and quantitative regression of water content, then developed a set of practical terahertz spectrum data processing and analysis softwar e.Introduced the terahertz time-domain spectrum system structure, working principles and important parameters were introduced. The model of terahertz optical parameter extraction based on Finel formula is derived, and the time domain spectrum is transformed into the frequency domain absorption coefficient spectrum and the refractive index spectrum, which is convenient for the quantitative analysis of the material.In view of the problem that the frequency resolution is reduced due to the length truncation of the terahertz time domain signal, the terahertz time domain reflection peak removal method is deeply studied. Due to reflection in the terahertz time-domain spectrum system devices, there are multiple reflection peaks in terahertz time-domain signal. Truncating these reflection peaks would cause a too short signal, resulting in reducing the frequency domain resolution, which is the main factor that affects the identification of the material. Therefore, the time domain repair technology, the principle of adaptive noise cancellation and empirical mode decomposition of three kinds of methods to remove the reflection peaks, so as to improve the frequency resolution. Compared with the former two methods, the third method not only removes the reflection peak in time domain better, but also improves the information retention capacity of the absorption spectrum, which lays a foundation for subsequent material quantitative analysis.In the process of terahertz experiment, water vapor content in the air has a great influence on the measured results, so it is necessary to accurately obtain and control experimental environment humidity. Test of terahertz time domain signals under different humidities was implemented, and on the basis of absorption spectrum the principal component analysis was employed to reduce the number of data dimension. Then monadic linear regression and partial least-squares regression was employed for quantitative analysis of the percentage of water vapor in the air. The experimental results show that the two methods can effectively identify the current environment humidity.Combined with the application, a terahertz time-domain spectrum data processing and analysis software based on MATLAB GUI was developed. The software can process terahertz time-domain spectrum data in batch, and provide visual observation and storage graph and the results of quantitative analysis, which indicates the characteristics of high efficiency and usage convenience.