On-site Rapid Determination of Content of Glycerol Triacetate in Cigarette Filter

KeyWord:cigarette filter  ?glycerol triacetate  ?hand-held NIR  ?PSO£­ELM

Abstract:

      Glycerol triacetate is used as the plasticizer and hardener in molding process of cigarette filter£¬its content will not only affect the rigidity£¬resistance and filtration performance of a cigarette filter£¬but also have an impact on the perceived quality of cigarette smoke£®In order to improve the efficiency for detection of glycerol triacetate content£¬a novel hand-held near-infrared spectroscopy£¨NIR£© technique combined with particle swarm optimization£­extreme learning machine£¨PSO£­ELM£© algorithm was applied to build a prediction model for glycerol triacetate content£®Meanwhile£¬it was also compared with the traditional partial least squares£¨PLS£© model and extreme learning machine£¨ELM£© model£®The experimental results showed that compared with PLSR model and ELMR model£¬the PSO£­ELM model was much better with a determination coefficient£¨r2£© of 0.921 2£¬which was higher than 0.860 4 for PLSR prediction model and 0.877 2 for ELMR prediction model£®Besides£¬the root mean square error of prediction£¨RMSEP£© for PSO£­ELM model was 0.3921 2£¬which was smaller than 0.497 72 for PLSR model and 0.470 18 for ELMR model£®The above results indicated that the established model could be applied to the rapid and accurate determination of the content of glycerol triacetate in cigarette filter£¬which provided a good reference for on-site rapid determination on the content of glycerol triacetate£®