One of the most common methods of damage detection in structures is modeling the finite element of that structure along with Newman and Drickel conditions and matching it to the experimental vibration tests performed on the same structure.The main goal of updating the model is to reduce the difference between the data obtained from the experimental vibration test (performed) and the analytical model of the finite elements. From a practical point of view, updating the model combines the skills of the numerical analyst and vibration test engineer in one place and requires the application of advanced estimation and calculation techniques for the desired corrections.This work is done by correcting the physical parameters, geometry or boundary conditions of the initial finite element model, relying on the experimental test results such as the frequencies and shape of the measured modes or the frequency response functions.The non-compliance mentioned is due to errors such as the discretization process error of the finite elements, the modeling error of the connections and boundary constraints, the parametric error resulting from the geometrical characteristics and physical properties, the error of the governing equations, the measurement and signal processing errors, the presence of noise and information defects caused by Pointed imperfect measurements are collectively known as system uncertainties. Considering the inherent numerical characteristic of the Bayesian method on reducing the adverse effects of uncertainty, this article updates the finite element model in the condition of incomplete measurement and using the frequency response function on a 40-member truss in order to detect failures resulting from the reduction of stiffness and has committed a crime The current method is written with Bayesian formulation, but Metropolis algorithm is used in repetition and reaching the updating model.The results show the improvement of speed and accuracy in the process of identifying structural failures.