An Adaptive, Multi-Resolution, Modal Parameter Extraction Technique

Materials subjected to random loading can deteriorate in a complex fashion. A technique for monitoring the manner in which this decay occurs can be useful, especially, for comparative analysis. Oftentimes, estimates of a system’s condition are made using estimates of its modal parameters, particularly natural frequency. This paper presents a new technique by which stochastic excitation and response data are used to continually determine the system’s instantaneous impulse response function, which is then used to provide continuous estimates of the system’s natural frequency. This is achieved using a digital finite-impulse-response (FIR) filter in conjunction with a normalized least-mean-squared (NLMS) technique. Results obtained from the analysis of physical and numerically simulated single degree-of-freedom (SDOF) systems subjected to random base excitation are presented. The results show that the new technique is eminently capable in tracking variations in the natural frequency of SDOF systems, with fine temporal and spectral resolution.