Application of SVM in Gas Sensor: Quantitative Analysis of CO2

Objecttive: According to the difficult in selecting parameter of SVM when modeling on the gas quantitative analysis, and existing methods need long time to finish, SVM optimized by improved grid search method was proposed to built a model to quantitatively analyse infrared spectrum of CO2 gas. Methods: We analyze 15 samples of CO2 gas at concentrations ranging from 500 ppm to 18% based on SVM. According to this method, the spectrum data of CO2 is optimized. The kernel function leads SVM and calculate the concentration. By using improved grid search, quantitatively analyzed 15 different concentrations of CO2 in the range between 500 ppm similar to 18%. Results: The experiment results show that this method gets c=1.412, g=0.25. And the prediction error is less than 5%. Conclusion: And method of grid search combined with SVM has a certain potential for development and mining space in gas quantitative analysis modeling in the infrared spectrum of CO2 gas.