In recent time the research in the field of mechanical engineering in focused towards the development of advanced time-optimized numerical methods. For this reason, the modeling approaches within structural dynamics are orientated towards the use of modern substructuring techniques. According to the presented issues, the tendency in the field of experimental dynamic substructuring is to develop a more efficient, robust and practical methods. The majority of the development is therefore orientated towards the development of a hybrid-substructuring methods. This kind of methods combine efficiency and robustness of the numerical dynamics substructuring methods and the real-system properties though the experimental dynamics substructuring methods. In this thesis the extension of well-established hybrid method called System Equivalent Model Mixing in presented. By developing the criterions for identification of so called bad measurements the major improvement in the sense of applicability and robustness of the method is presented. The method enables the reconstruction of frequency response functions associated with bad measurements using the expansion techniques by considering the consistency of numerical model. Finally, the comparison of developed method is presented with basic system equivalent model mixing formulation by coupling two identical beams.