Dynamical Adsorption Behavior Prediction of Dried Tobacco Leaf Heterogeneous Interfaces through Simulation and Image Recognition Techniques.

The process of spraying water and flavorings on dry tobacco is an important factor in the industrial environment and product quality. Tobacco as a complex porous fiber material, the interfacial transfer process of water is complex. In this study, machine learning and image recognition techniques were utilized to quickly obtain the structural parameters of the tobacco surface and construct a cellular structure model of the tobacco surface. In situ observation of the droplet impact spreading process was carried out using a high-speed camera to explore the droplet dissipation dynamics on different surfaces. And the competing processes of droplet wetting and evaporation under the influence of surface microstructure were determined by combining experimental studies and finite element simulation calculations. Based on the characteristics of tobacco pore size distribution, the infiltration under gas-liquid two-phase action was transformed into single-phase flow transfer under capillary force, and the continuous droplet infiltration process was simulated. A parallel artificial membrane permeability measurement method of bionic tobacco waxy layer was constructed for the screening of spray dosing copenetrant. This study brings new insights into the wetting of porous fibrous materials and is important for exploring the wetting process and additive development process influenced by the microstructure.

Study of metabolite differences of flue-cured tobacco from Canada (CT157) and Yunnan (Yunyan 87).

In order to comprehend the dissimilarities in tobacco quality between Canada and Yunnan, a comparison of the aroma components was conducted using GC-MS and HPLC analysis, coupled with orthogonal partial least squares discriminant analysis (OPLS-DA). The study revealed the detection of a total of 81 aroma components and 22 non-volatile components in both varieties of tobacco leaves. Specifically, there were 102 components of Canada tobacco leaves and 103 components of Yunnan tobacco leaves. Subsequently, a screening was performed on these two types of tobacco leaves, identifying 51 differential components, which accounted for approximately 49.5 % of the overall components detected. Among these, Canada tobacco exhibited a higher concentration of 22 components, comprising roughly 36.4 % of the total, which were primarily composed of semi-volatile organic acids and sesquiterpenes. On the other hand, Yunnan tobacco was characterized by a comparatively higher content of 43 components, constituting approximately 63.6 %, including fatty acid esters, phenols, diterpenes, sugars, and amino acids. Comparatively, Canada tobacco demonstrated elevated levels of fatty acids and sesquiterpenes, while the content of fatty acid esters and diterpenes was relatively lower. These distinctions in aroma components potentially contribute to the varied sensory aroma profiles exhibited by the two types of tobacco.

Across different instruments about tobacco quantitative analysis model of NIR spectroscopy based on transfer learning.

With the development of near-infrared (NIR) spectroscopy, various calibration transfer algorithms have been proposed, but such algorithms are often based on the same distribution of samples. In machine learning, calibration transfer between types of samples can be achieved using transfer learning and does not need many samples. This paper proposed an instance transfer learning algorithm based on boosted weighted extreme learning machine (weighted ELM) to construct NIR quantitative analysis models based on different instruments for tobacco in practical production. The support vector machine (SVM), weighted ELM, and weighted ELM-AdaBoost models were compared after the spectral data were preprocessed by standard normal variate (SNV) and principal component analysis (PCA), and then the weighted ELM-TrAdaBoost model was built using data from the other domain to realize the transfer from different source domains to the target domain. The coefficient of determination of prediction (R 2) of the weighted ELM-TrAdaBoost model of four target components (nicotine, Cl, K, and total nitrogen) reached 0.9426, 0.8147, 0.7548, and 0.6980. The results demonstrated the superiority of ensemble learning and the source domain samples for model construction, improving the models' generalization ability and prediction performance. This is not a bad approach when modeling with small sample sizes and has the advantage of fast learning.

Quality evaluation of flue-cured tobacco by near infrared spectroscopy and spectral similarity method.

Herein we propose near infrared (NIR) spectroscopy as a rapid method of evaluating the quality of agricultural products. Unlike existing quantitative or qualitative models, quality similarity is characterised using spectral similarity. Key factors of the spectral similarity method were investigated, including variable selection, pre-processing and similarity measures. Sophisticated techniques were developed to ensure the reliability of similarity algorithm. The proposed method was tested by quality similarity of flue-cured tobacco samples. The results demonstrated that the quality-related factors between the target and the similar samples (determined by spectral similarity), showed high similarities. This new method has the potential to characterise product quality effectively and could be a useful new alternative to the widely used PLS models.