The spectral changes in a vibration might be useful information for health monitoring or diagnostics on structural constructions. However, it is difficult to estimate the structural transfer function (TF) using the external vibration-source signals. The authors have been investigating cumulative harmonic analysis (CHA) for estimating the resonance frequencies of the TFs hidden in external random-source vibrations. CHA emphasizes the poles of the TF by using the spectral accumulation of the observed response record on a sample-by-sample basis. Thus, CHA estimates and visualizes the system-damping conditions without needing TF measurements using a particular testing signal. A scale-model experiment confirmed that the TF conditions could be well visualized by using a modal analysis that includes a CHA. The effects of fluctuations in the external source signals on the spectral analysis by using a CHA will be described in this paper by introducing a cepstral analysis. This might create an important problem in not only CHA but also the conventional statistical energy analysis (SEA). In particular, the spectral resolution of an adjacent pair of poles will be discussed from the viewpoint of the cepstral fluctuations in the observed response records. Cumulative harmonic analysis, Structural health monitoring, Diagnostics, Transfer function, Acoustic signal processing.
展开▼