Infrared spectrum analysis of the gas in coal mine based on SVM

Gas detecting in the coal mine is always a significant problem. As the molecule of the gas may absorb the light at some wavelength, analysis on the gas absorbing infrared spectrum can be made for contributing to the gas concentration. On the other hand, coal gas has diversified composition and large concentration range, and the characteristic absorbing spectrum line of the composition overlaps each other, while support vector machine is a kind of machine learning method based on statistical learning theory, and it is mainly useful for small samples, therefore, in this application support vector machine is associated with infrared spectrum analysis to investigate the coal gas. The research concludes the establishment of infrared spectrum data sample, the training and testing on SVM calibration model, the implement of SVM calibration model. The results show that the precision and reliability can satisfy the requirements detecting the gas concentration. The precision and sensitivity is high, the response speed is fast, and gas can be analyzed on line continuously. The proposed approach is helpful and practical.