To enhance the accuracy of finite element model simulation, a model updating method based on Bayesian theory is proposed, and the updating efficiency is improved by combining improved Markov Chain Monte Carlo (MCMC) algorithm and surrogate model. A Radial Basis Function (RBF) surrogate model is constructed using the parameters to be updated as inputs and the finite element model modal responses as outputs. Whale Optimization Algorithm (WOA) is introduced into the MCMC algorithm and the uncertain parameters are updated. Finally, a numerical study on a simply supported beam and an experimental study on a three-story steel frame are conducted to verify the accuracy of the proposed method. The results show that WOA can significantly improve the stability and convergence speed of MCMC algorithm, the updating efficiency can be improved by 13.9% at most, and the maximum frequency errors of the simply supported beam model and the three-story steel frame model updated by WO-MH algorithm are 0.009% and 2.41%, respectively. The proposed model updating method can effectively enhance the simulation accuracy of the finite element model under both two-dimensional and eight-dimensional inputs, which provides technical reference for lean simulation and optimal design of building structures.