There are many influence factors in the precision and accuracy of the quantitative analysis with LIBS technology.According to approximately the same characteristics trend of background spectrum and characteristic spectrum along with the change of temperature through in-depth analysis,signal-to-background ratio(S/B) measurement and regression analysis could compensate the spectral line intensity changes caused by system parameters such as laser power,spectral efficiency of receiving.Because the measurement dates were limited and nonlinear,we used support vector machine(SVM) for regression algorithm.The experimental results showed that the method could improve the stability and the accuracy of quantitative analysis of LIBS,and the relative standard deviation and average relative error of test set respectively were 4.7% and 9.5%.Data fitting method based on signal-to-background ratio(S/B) is Less susceptible to matrix elements and background spectrum etc,and provides data processing reference for real-time online LIBS quantitative analysis technology.