Aiming at the non-linear standard curve of the relationship between concentration and light intensity in the ICP-AES analysis process, a standard curve fitting method based on the OLS algorithm and the improved LM algorithm is proposed to realize the accurate analysis of element concentration. Quadratic polynomial, cubic polynomial, limbek expression and Lwin expression are used as the nonlinear model of the standard curve to suit the distribution characteristics of different data. According to the loss function characteristics of each expression in the fitting process, the OLS algorithm is used to calculate the optimal fitting parameters of quadratic and cubic polynomials, and the improved LM algorithm is used to obtain the optimal fitting parameters of limbek and Lwin expressions, so as to realize the fitting of standard curve. The light intensity measured data of a series of standard samples with different concentrations of Sb, Cd, Sn, Mo Ni and Ba elements are used to carry out the standard curve fitting experiment of the above method. The experimental results show that the concentration-light intensity data of each element can be fitted to obtain a nonlinear standard curve with a R 2 above 0. 999, and the relative error of the fitted concentration of the standard curve to the known data points is within ±5%.