Over the last few decades, there has been great interest in the development of a structural health moni-toring (SHM) methodology based on vibration data. The quantity and quality of the measured data, i.e., the number of sensors and the corresponding locations are very important for the success of SHM utilizing measured dynamic re-sponses. In order to extract the most information from the measured data and reduce the uncertainties of the identified model parameters, a methodology of optimal sensor configuration for structural model parameters identification is pre-sented. In order to avoid the influence of modeling error induced by traditional finite element method based on static shape function on the results of structural dynamic characteristics and optimal sensor placement, spectral finite element method being a dynamic modeling method with high-accuracy is employed to model the target structure in the proposed methodology. In addition, minimum of the uncertainties in model parameter estimates is taken as the optimality criterion for placing sensors, and information entropy measure is used to quantify these uncertainties which are calculated by the Bayesian statistical identification methodology. The information entropy measure is minimized over the set of possible sensor configurations to optimally locate a given number of sensors by an integer-coded genetic algorithm. Both nu-merical simulation and laboratory experiment are carried out for a periodic pipe-beam model with flexible joints on elastic foundations to verify the proposed methodology.