New Chemometric Algorithm In Harmful Component Analysis Of Cigarettes

There is a growing concern on Smoking and Health, especially the harmfulcomponents in the cigarette smoke. Currently, the tobacco industry on the detection ofharmful components in the cigarette smoke mainly depends on a variety of expensiveequipment, such as, GC-MS、 LC-MS or more complex instruments. Samplepretreatment of these methods are complicated, require skilled technicians, and existenvironmental pollution problems. We consider the use of infrared spectroscopy toanalyze the harmful components in the cigarette smoke. Because techniques based oninfrared spectrum are cost-effective, easy to curry out without expensive equipmentand chemical reagents, without the limit of the sample morphology, samplenon-destructive, rapid analysis, and provide a wealth of information, etc. Consequently,we combined the infrared spectroscopy and chemometric methods for rapid analysis ofsome of the harmful ingredients in cigarettes.This paper mainly deals with the following aspects:1. In this research work, considering the characteristics of infrared spectroscopy,we put forward a new algorithm, adaptively configured wavelet packet transformpartial least squares as globally optimized by particle swarm optimization algorithms(DSPO-WPT-PLS) for IR spectra analysis. WPT has been proved to be an effectivemethod which works well in extracting feature information from raw IR spectra. Afterfeature spectra extracting and denoising using WPT, partial least squares regression,which is invoked for calibration model building. The two steps of IR spectroscopyanalysis as mentioned above are closely related to each other via a synergeticoptimization by using a PSO algorithm to simultaneously determine the optimalwavelet packet decomposition tree for WPT, useful wavelet coefficients and hiddenvariables for PLS modeling. So build the new algorithm for IR spectroscopy analysis.In the second chapter of this article, determination of the content of benzo[a]pyrenewith GC-MS, wich is a harmful component in cigarette mainstream smoke. Thencombines the new algorithm and attenuated total reflection infrared spectroscopy(ATR-IR), and establishing a quantitative model for analysising harmfulbenzo[a]pyrene, the result of this method is good. We also use the traditional method,such as MLR、 PLS、 and based on stepwise regression variable selection PLS,analyze the harmful ingredients in the mainstream smoke of benzo[a]pyrene. The results of different methods are compared, indicating DPSO-WPT-PLS is more robustand reliable than MLR、PLS、and based on stepwise regression variable selection PLS.Demonstrate that the new algorithm can be applied to the analysis of harmfulingredients of benzo[a]pyrene in cigarette mainstream smoke. And expected to becomea new effective, rapid quantitative analysis methods in analysising harmful ingredientsof benzo[a]pyrene in cigarette mainstream smoke.2. In the third chapter of this article，we use micro oxygen bomb calorimeterdetermination of the value of tobacco combustion heat, then combines the newalgorithm and ATR-IR, and establishing a quantitative model for analysising tobaccocombustion heat. In order to optimize the conditions for the analysis of tobaccocombustion heat. We studied a different mother wavelet and waveletpacket decomposition level influence on the results. Based on the result, we select‘db4’ as mother wavelet, and the wavelet packet decomposition level is four. The resultof this method is good. And compare with traditional methods, such as PLS、based onstepwise regression variable selection PLS, the results of different methods arecompared, indicating DPSO-WPT-PLS is more robust and reliable than PLS and basedon stepwise regression variable selection PLS. And expected to become a new rapid,effective quantitative analysis way for tobacco combustion heat.3. In the fourth chapter of this article, determination of the content of the harmfulcomponents of four tobacco-specific nitrosamines in mainstream smoke withLC-MS-MS. The four tobacco-specific nitrosamines are NNN, NNK, NAT, NAB. Thencombines the new algorithm and ATR-IR, and establishing a quantitative model foranalysising harmful components of four tobacco-specific nitrosamines in mainstreamsmoke. Optimized modeling parameters, DSPO-WPT-PLS simultaneous analysis offour nitrosamines, have achieved satisfactory results. We also compare with thetraditional PLS、based on stepwise regression variable selection PLS, results show thatthe DSPO-WPT-PLS is superior to traditional PLS, the model is more robust, and themethod is feasible. This analysis method contributes a new rapid, effective analysisway for the harmful components of tobacco-specific nitrosamines in mainstreamsmoke.