NIR Detection of Alcohol Content Based on GA-PLS

摘要

Alcohol content is an important indicator of many products, rapid and accurate analysis is the key link of commodity inspection and productive process. Study using NIRS model to detect alcohol content of wine, adopted genetic algorithm partial least-squares (GA-PLS) method to analyze the nearinfrared spectroscopy (NIRS) characteristic wavelengths of alcohol content, the best NIR GA-PLS model is established. Experiment find GA-PLS can flexible and effective select out the characteristic wavelengths, can not only get rid of the useless information wavelengths, but also improve model's predicted precision, therefore, the predicted precision of GA-PLS model superior to PLS model established with global spectrum. The best predicted effect is obtained when 133 wavelengths with higher selected frequency join modeling, its root mean square error of prediction (RMSEP) is 0.0066 and correlation coefficient of prediction (Rp) is 0.9996. The results show, NIRS combined with GA-PLS method can detect alcohol content of wine rapidly and accurately, expected to achieve rapid detection of alcohol content online.