Abstract: It has been challenging to correctly separate the mixed signals into source components when the source number is not known a priori. To reveal the complexity of the measured vibration signals, and provide the priori information for blind source separation, a novel source number estimation method is proposed based on independent component analysis (ICA) and clustering evaluation analysis. The performances of the proposed method and the information-based source number estimation methods are comparatively studied via numerical case study and experimental study on a thin shell structure, and the results show that the proposed method can correctly and reliably estimate the source number from the modulated vibration signals, and thus has more powerful performances in engineering applications.

Key words: source number estimation;independent component analysis;clustering evaluation;source separation;shell structure