RÉSUMÉ
Whennearinfraredspectroscopyisappliedtoon-linemonitoringandcontroloftobaccoflavors,the variation in temperature can severely deteriorate the predictive performance of near infrared spectroscopic calibration models and results in a significant increases of the root mean square error value for the main constituents in syrup samples from 2. 4% to 29. 0%. In this paper, near infrared spectroscopy has been incorporated with an advanced calibration transfer method-loading space standardization to effectively eliminate the deteriorate effects of temperature variation on quantitative results and finally realize the fast and accurate on-line quantitative monitoring and control of tobacco flavors. The root mean square error value for the main constituents in syrup samples is successfully retained at a satisfying low level of 3 . 8%. The results of this paper will provide technical support for the preparation, preservation and use of tobacco flavors, and realize on-line process quality control of cigarettes.