CONCLUSIONS Frequency based decoupling techniques allow for component identification without ever taking measurements on the individual component. The Impedance, Mobility and Inverse FBS decoupling techniques were studied in this paper. Analytical and experimental models were used to determine the robustness of each technique, and attempt to address issues common with experimentally obtained FRF. All techniques produce accurate results when using pure analytical FRF, however, the Inverse FBS technique requires measurements to be on both components of the system, on opposing sides of each connection. Obtaining this measurement is very difficult if not impossible in many practical applications. When noise is introduced to the FRF, all techniques amplify the noise due to the matrix inversions. This amplification is significantly higher with the Inverse FBS technique as multiple inversions are performed. Impedance and Mobility are especially advantageous when internal DOF are available, as this was shown to greatly improve the accuracy analytically and experimentally. For the beam models studied in this paper, rotational DOF were address by using two translational DOF to approximate rotational DOF. While this method was shown to be a valid approximation, the rotational coupling stiffness in this model was very low and decoupling was shown to be accurate without including these DOF. Without including internal DOF, experimental model AB-TR produced more accurate results than AB-TT. However, including two internal DOF with experimental model AB-TT significantly improved the results and provided the most accurate experimental results. RECOMMENDATIONS Promising results have been produced using the Mobility technique with experimentally obtained FRF. However, this structure was measured in a way to represent a planar structure with only two DOF per node (translational and rotational). The connection stiffness was relatively small for the rotary stiffness when compared to the translational stiffness. Future work should address higher coupling stiffness. Also, cases should be run to address relative subcomponent size. The structure in this work was symmetrical in which modes of each of the components are approximately at the same frequencies. Future work should address the sensitivity of the Mobility approach to relative component mass and/or stiffness. 184
RkJQdWJsaXNoZXIy MTMzNzEzMQ==