Wavelength Selection for Prediction of Polyphenol Content in Inward Tea Leaves Using NIR

Abstract

In this work total polyphenol contents in tea leaves have been estimated by the near infrared reflectance (NIR) spectroscopy and partial least squares (PLS) algorithm. During sample acquisition the number of variable is quite high for each spectra and whole range of spectra may not play an important role for building the calibration model of PLS algorithm. Selection of proper region for a particular application is an important task. Here, optimum wavelength was determined by genetic algorithm (GA) and particle swarm optimization (PSO). PLS algorithm was used to produce the fitness curve of PSO and GA. Training and testing was done by leave –one-sample out cross-validation during the model calibration. Testing and training was done using specific windows of wavelength. The optimum range was determined to be from 1027.75 nm to 1104.75 nm. The RMSECV value for the optimum range was observed to be 1.05.