Combining support vector machine with infrared spectrum analysis, the infrared spectra analysis model of mixed gas is established. Taken the mixed gas containing hydrocarbon as an example, the selection and optimization of model parameters is researched through experiment. Beginning with the SVM analysis model, infrared spectrometer and spectra data, spectra analysis band and spectrometer scanning interval, the detailed study of how the parameters like the types of kernel function for the SVM analysis model, penalty factor C, method of spectra data preprocessing, spectrometer scanning interval, spectra analysis band, etc. Affect the analysis result are carried out. The experimental results show that under the condition of the SVM analysis model is determined, a reasonable selection and optimization of the SVM analysis model parameter can improve the accuracy of the analysis results, and it has a practical application value.