84 S. Marwitz and V. Zabel 1 a b c 2 3 4 5 6 7 8 9 1011121314 0 2 4 6 8 10 12 14 Δf [%] 1 2 3 4 5 6 7 8 91011121314 0 1 2 3 4 5 6 7 8 ζ [%] PoSER PoGER PreGER 1 2 3 4 5 6 7 8 91011121314 0 20 40 60 80 100 MAC [%] Fig. 7.7 Manually selected modal parameters for 14 modes of vibration that were chosen. No bar indicates that the respective mode could not be identified. The comparison of the three merging strategies is based on the same experimental data. (a) Frequency differences in percent relative to the reference measurement. (b) Absolute damping values. Reference measurement is marked by a dashed horizontal line. (c) Modal Assurance Criterion of identified modeshapes with respect to the reference measurement In order to improve the results, several measures are conceivable. The placement of the scan heads relative to the structure could be optimized based on a preliminary experimental and/or numerical modal analysis. Additionally the aerodynamic excitation might not be the best choice to obtain a proper modal analysis. Therefore, different types of excitation could be chosen. As mentioned already the selection of suitable reference points for each scan head could also be based on preliminary and/or numerical studies to further improve the quality of the reference signals. And finally, there was no evaluation of adequacy of the chosen parameters for the data acquisition, the data processing, the stochastic subspace identification algorithm and the selection of stable poles. 7.5 Conclusion In the presented study, it was shown, that it is possible to conduct 3D SLDV measurements without any additional input or output reference signal. A post-estimation 3D coordinate transformation algorithm was developed and applied to experimental data. Three different merging strategies were evaluated in terms of their applicability and accuracy in the context of the reference-based, covariance-driven stochastic subspace identification method. In this study, the classical PoSER approach led to the best results, while the PreGER approach was not satisfying. The PoGER merging approach was found to be a good compromise between accuracy of results and time expenditure. The proposed 3D coordinate transformation algorithm was shown to be applicable, but prone to errors. In particular any inaccuracy that occurred during merging led to large errors after the 3D coordinate transformation. However, these deficiencies seem to be mainly related to the selection of suitable reference points. For further research it is recommended to evaluate the proposed approach using a different test structure and/or different modal analysis methods. Additionally, it should be generally possible to use two and more single SLDV to imitate commercial 3D SLDV. However, synchronized demodulation of the doppler signal and data acquisition must be tackled.
RkJQdWJsaXNoZXIy MTMzNzEzMQ==