One of the important applications of modal analysis is model updating finite element method by using frequency response functions (FRF) in order to damage detection in the structures. The method is a sensitivity-based model updating approach which utilizes a pseudo-linear sensitivity equation and is applied to identify the location and amount of the changes in structural parameters. The method is can be improved against the adverse effects of incomplete measurement, measurement errors and uncertainties such as modeling errors and noise effects. This paper tries to upgrade to solve the problem in a more suitable interval of the response with the Bayesian approach and repeated testing of multiple data resulting from the model updating method with frequency response functions. In fact, although statistical approach can bring a logical result in some problems, but in engineering problems, according to the state of the problem and the answer of repeated experiments (or in non-laboratory problems, according to Monte Carlo repetitions), it can be added conditions to the problem in which he achieved a better estimation of the results by using the Bayesian statistical method. The experimental setup consists of a free-free aluminum beam, where changes are introduced by reducing the stiffness and adding mass at certain parts of the beam. The results indicate that the location and the size of different level of changes in the structure can be properly identified by the method.