Recovery of operational deflection shapes from noise-corrupted measurement data from CSLDV: Comparison between polynomial and mode filtering approaches

Continuous Scanning LDV measurement methods allow estimates to be made of high spatial density vibration deflection shapes (ODSs) in very short testing times. The correct reconstruction of the ODS depends on the quality of the acquired data. Speckle and other types of noise, usually resulting from surface treatment of the specimen can be detrimental in this recovery process. This paper presents a comparison between two alternative methods for recovering ODSs from CSLDV data. The wellknown method is based on the fact that the amplitude modulated response can be processed by calculating the peak amplitudes and phases of the spectral sidebands and, eventually, the deflection shapes can be calculated by conversion of these peak amplitudes into polynomial coefficients. The second method exploits the idea that, if the mode shapes of the structure under test are known a priori, e.g. from a numerical model or from an analytical formulation, it is possible to settle a procedure that searches for similarities between those known mode shapes (the candidate mode shapes) and ODSs that actually modulate the CSLDV signal. This procedure can be considered a pattern matching technique that is able to identify the resonance frequency related to each ODS and the mode shapes that better match with ODSs excited making it possible to filter out the uncorrelated noise. Both methods have been applied to data acquired from measurements on the tailcone of a Lynx helicopter, which contained levels of noise disrupting the correct peak amplitudes data processing. Comparison with step-scan results was done in order to have a reference technique and to understand how optical noise issues could affect the measurement.