In order to accurately predict the springback of high-strength steel， it is necessary to use the material model parameters that accurately and comprehensively describe the material mechanical properties. Therefore， an inverse calculationmethod of the parameters of the Barlat89 and Yoshida-Uemori （Y-U） combined model was proposed based on the modified Levenberg Marquardt （LM） optimization algorithm. Using high-strength steelDP780 as an example， the mechanical property curve of the material was obtained through unidirectional tensile and tensile compression tests. LS-DYNA software was used to simulate and analyze the object of the test. The modified LM algorithm was used for continuously optimizing the material model parameters preset by simulation. Finally， the simulation output equals the material property curve obtained from the experiment in the sense of least squares， and the optimal material combination model parameters are obtained. The results show that the correlation coefficient of the LM algorithm is 0.951 4， and the convergence of the LM algorithm is good. The inverse Barlat89 and Y-U model parameters can accurately describe the uniaxial tensile and tensile compression mechanical properties of DP780 materials at the same time. The simulation curve fits well with the test curve， and the average relative error of the two curves is 4.65%. The material model parameters obtained by this method reflect the mechanical properties of materials under forward and reverse loading and can markedly improve the prediction accuracy of springback.