A novel damage detection method based on flexibility identification theory and data fusion technique

An improved flexibility-based method is proposed in this study for damage detection. In this method, multi-scale convolution was utilized to decrease the interference of the measurement noise and the Dempster-Shafer evidence theory was adopted to combine all scale information to amplify the damage characteristics. Three main features distinguish the proposed method from the previous studies: 1) It is a kind of no-baseline flexibilitybased method; that is, this method can locate the damage with the absence of intact structural flexibility serving as baseline; 2) Flexibility is estimated without requiring to know the structural mass, which is necessary in the traditional method for flexibility estimation; 3) By utilizing multi-scale space theory and data fusion approach, the proposed method has superior noise tolerance ability. Examples of both numerical and experimental types were studied to reveal the effectiveness and accuracy of the proposed method in different noise levels. Comparison between the traditional method and the proposed method demonstrated that the latter was quite suitable to detect damage to beams structure in a noisy environment.