Abstract | ||
---|---|---|
The main chemical compositions of tobacco is one of the significant factors to decide the quality of tobacco, and it is time consuming, expensive, unrepeatable and destructive to obtain the chemical compositions in laboratories. This paper investigates the relationship between tobacco near infrared (NIR) spectra and chemical composition, thus the chemical composition can be easily obtained through the NIR spectroscopy that is a quick, convenient, accurate and low-cost technique. An intelligent method is proposed based on least squares support vector machines (LS-SVM), and a method based on partial least squares regression (PLS) is also proposed comparative study. The obtained results show that the proposed methods are effective and feasible. The best prediction accuracy for the seven different chemical compositions is obtained by the LS-SVM method, which can reach an accuracy at 99.85% for the 400 training samples and 97.92% for the 100 testing samples. |
Year | DOI | Venue |
---|---|---|
2017 | 10.1109/ICInfA.2017.8078954 | 2017 IEEE International Conference on Information and Automation (ICIA) |
Keywords | Field | DocType |
chemical composition,least squares support vector machines,partial least squares regression,near infrared spectroscopy | Least squares,Chemical composition,Biological system,Computer science,Control theory,Near-infrared spectroscopy,Partial least squares regression,Support vector machine,Spectral line,Spectroscopy,Linear regression | Conference |
ISBN | Citations | PageRank |
978-1-5386-3155-3 | 0 | 0.34 |
References | Authors | |
5 | 4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Di Wang | 1 | 1337 | 143.48 |
Fengchun Tian | 2 | 27 | 10.47 |
Simon X. Yang | 3 | 1029 | 124.34 |
Zhiqin Zhu | 4 | 128 | 14.67 |