Ground vibration test results for modal updating of aircraft

Computational dynamic models are developed during design stage of aircraft. These models are used for preliminary assessment of structural load levels and controllability of spacecraft in orbit. They are also necessary to ensure structural strength and aeroelastic stability of aerospace vehicles. The computational models, which are built on technical documentation, are updated through ground-based verification of spacecraft and experimental modal analysis of aircraft. Methods for updating computational models are divided into stochastic and deterministic ones. In the present work the problem of obtaining input data for the deterministic updating method is solved. The method minimizes the objective function defined as the sum of squared differences between experimental and computational data. It is assumed that the computational dynamic model of aircraft is based on the free vibration equations. That is why inertia and stiffness matrices are to be updated based on experimental data, such as generalized masses and natural frequencies. Since damping forces are not included in the free vibration equations, a monophase oscillation method is used for ground vibration testing. That method does not require prior identification of the dissipative properties of the dynamic system and allows independent determination of the mass–stiffness characteristics of the test object, regardless of damping properties. Test modes for determination of eigenfrequencies, eigenmodes and generalized masses are described. The reliability of experimentally determined modal parameters has been investigated in order to determine their applicability as target parameters for modal updating. The errors in modal results caused by random measurement errors of vibration amplitudes and by the interaction of modes with closely spaced natural frequencies have been evaluated. It is noted that errors in determining natural frequencies using the phase resonance method are an order of magnitude lower than errors in measuring vibration amplitudes. At the same time, errors in estimating generalized masses using known methods are an order of magnitude higher than those in natural frequencies. The interaction of modes with closely spaced natural frequencies demonstrates itself in shifts of phase resonance frequencies and errors in determining generalized masses. For example, errors in estimating natural frequencies using phase resonance do not exceed 1 % over a wide range of parameters for closely spaced modes. Meanwhile, determining generalized masses with an error of 5% is only possible within a narrow range of these parameters. As a result of the conducted research, it has been established that the reliability of experimental estimation of natural frequencies justifies their use as parameters of the objective function for updating the stiffness matrix of the computational model. At the same time, updating the inertia matrix developed at the design stage is impractical due to the large errors in estimating generalized masses.