Component Spectrum Recognition for Mixed Gas Based on SVM

Abstract:

Article Preview

As for the problem that component gas characteristic spectrum lines overlaps seriously in the identification of Mixed Gas, Support Vector Machine is introduced for the identification, and an one-by-one identification methods for Mixed Gas classification based on the binary category identification model based on the support vector machine is proposed in this article. One-by-one category identification is carried out for each mixed gas when the characteristic spectrum lines are overlapped seriously and is transformed in high dimensional space into linear by SVM kernel function transformation. In the experiment for gas component identification of a natural gas, we compare the recognition results affected by different kernel functions, data preprocessing, feature extraction, numbers of training samples and other conditions. The results show that the method has the correct recognition rate of over 97% for the natural gas whose concentration is over 1%, and it has a great promotional value both in theory and practical application.

Info:

Periodical:

Edited by:

Zhixiang Hou

Pages:

557-560

DOI:

10.4028/www.scientific.net/AMM.128-129.557

Citation:

P. Bai et al., "Component Spectrum Recognition for Mixed Gas Based on SVM", Applied Mechanics and Materials, Vols. 128-129, pp. 557-560, 2012

Online since:

October 2011

Export:

Price:

$35.00

[1] Vapnik V. Statistical Learning Theory. New York. John Wilev & Sonc. Inc., (1998).

[2] V. Vapnik. The Nature of Statistical Learning. New York: Springer, (1995).

[3] V. Vapnik. An Overview of Statistical Learning Theory. IEEE Transon NeuralNetWork[J], 1999, 10(5):988-999.

[4] Gradner J W, Brtlett P N. A brief history of electronic nose. Sensors and Actuators. 1994, (18-19): 211-220.

[5] Chih-WeiHsu, Chih-JenLin. A Comparison of methods for multiclass support vector machines. IEEE Trans. on NeuralNetworks. 2002, 13(2): 415~425.

DOI: 10.1109/72.991427

[6] Platt J, Cristianini N, Taylor J. Large margin DAG's for multiclass classification. Advances in Neural Information Processing Systems. Cambridge, USA: MIT Press, 2000, 547~553.

In order to see related information, you need to Login.