Wavelength Selection Method Based on Differential Evolution for Precise Quantitative Analysis Using Terahertz Time-Domain Spectroscopy

摘要

>Quantitative analysis of component mixtures is an important application of terahertz time-domain spectroscopy (THz-TDS) and has attracted broad interest in recent research. Although the accuracy of quantitative analysis using THz-TDS is affected by a host of factors, wavelength selection from the sample’s THz absorption spectrum is the most crucial component. The raw spectrum consists of signals from the sample and scattering and other random disturbances that can critically influence the quantitative accuracy. For precise quantitative analysis using THz-TDS, the signal from the sample needs to be retained while the scattering and other noise sources are eliminated. In this paper, a novel wavelength selection method based on differential evolution (DE) is investigated. By performing quantitative experiments on a series of binary amino acid mixtures using THz-TDS, we demonstrate the efficacy of the DE-based wavelength selection method, which yields an error rate below 5%.