Validation of Measured Dynamic Data Using Rigid Body Response

As multiple axis vibration testing has become more widespread, it has become increasingly important to ensure the instrumentation is accurately portrayed in the instrumentation table. However, errors do occur. The method used in this paper to help uncover these errors is based on the condition that at low frequencies (below any resonant frequencies of the object being studied) the response is essentially rigid body. The spectral density matrix (SDM) at a low frequency, of many more than six response measurements, is decomposed using singular value decomposition (SVD). Under the assumption of rigid body response, it is assumed that the first six singular vectors are linear combinations of the six rigid body modes. The best linear fit is then calculated for this fit. The measurements are then removed one at a time, and the reduction in the fit error is calculated. It is assumed that if the removal of a measurement reduces the error significantly, that measurement is likely in error.