Near infrared spectroscopy (NIRS) combined with a variety of algorithms was used to conduct rapid and non-destructive quality evaluation of Yuan’an yellow tea.The near-infrared spectra of 90 samples of Yuan’an yellow tea were obtained by scanning,and 11 different methods were used to preprocess the spectra to remove part of the noise information.Then the backward interval partial least squares (BiPLS) method was used to screen the characteristic spectral interval reflecting the quality of the sample and the genetic algorithm (GA) was used to accurately extract the characteristic spectral wavelength.Five NIRS prediction models of yellow tea quality were established.Finally,an attempt was made to analyze information of the spectral functional group.The results showed that the best spectral pretreatment method was multiple scattering correction.The characteristic spectral interval screened with BiPLS method was 9 003.2-7 497.9 cm-1,6 101.7-5 449.8 cm-1 and 4 601.3-4 246.5 cm-1,respectively.GA algorithm accurately screened 75 characteristic spectral data points.The Bi-GA-PLS model had the best robustness and predicted the quality of unknown samples (R2=0.951,RMSEP=1.57,RPD=5.27),which was preliminarily realized to quickly and accurately forecast the quality of Yuan’an yellow tea.By analyzing the spectral information,45 spectral data points mainly reflected the information of functional group including —CHx,C=O and —NHx,representing the monosaccharide,caffeine,theanine and free protein closely related to the quality of the samples.The other 30 spectral data points mainly reflected the information of the functional group including O—H,amide bond and C—H and C—C stretching,representing lignin,starch and cellulose.