Data fusion of multi-scale representations for structural damage detection

•A novel hybrid methodology is proposed to achieve the detection of multiple slight damage.•The method has a superior noise tolerant ability and damage sensitivity.•The D-S evidence theory is utilized to search the damage feature across all scales.•The presented methodology only needs a single order modal data, and do not require baseline data.

Despite extensive researches into structural health monitoring (SHM) in the past decades, there are few methods that can detect multiple slight damage in noisy environments. Here, we introduce a new hybrid method that utilizes multi-scale space theory and data fusion approach for multiple damage detection in beams and plates. A cascade filtering approach provides multi-scale space for noisy mode shapes and filters the fluctuations caused by measurement noise. In multi-scale space, a series of amplification and data fusion algorithms are utilized to search the damage features across all possible scales. We verify the effectiveness of the method by numerical simulation using damaged beams and plates with various types of boundary conditions. Monte Carlo simulations are conducted to illustrate the effectiveness and noise immunity of the proposed method. The applicability is further validated via laboratory cases studies focusing on different damage scenarios. Both results demonstrate that the proposed method has a superior noise tolerant ability, as well as damage sensitivity, without knowing material properties or boundary conditions.

Graphical abstractDownload high-res image (161KB)Download full-size image