Cepstrum-based damage identification in structures with progressive damage. (January 2019)

• Record Type: Journal Article
• Authors: Dackermann, Ulrike; Smith, Wade A; Alamdari, Mehrisadat Makki; Li, Jianchun; Randall, Robert B
• Other Names: Nguyen Andy guest-editor.; Chan Tommy HT guest-editor.; Zhu Xinqun guest-editor.
• Is Part Of: Structural health monitoring. Volume 18:Number 1(2019)
• Page Start: 87
• Page End: 102
• Publication Date: 2019-01
• Subjects: Structural health monitoring -- frequency response function -- operational modal analysis -- cepstrum analysis -- artificial neural networks -- damage identification -- progressive damage -- timber structure; Structural health monitoring -- Periodicals; Structural stability -- Periodicals; Strength of materials -- Periodicals; Nondestructive testing -- Periodicals; Constructions -- Stabilité -- Périodiques; Résistance des matériaux -- Périodiques; Contrôle non destructif -- Périodiques; Electronic journals; 624.17
• DOI: 10.1177/1475921718804730
• Languages: English
• ISSNs: 1475-9217

Abstract

This article aims at developing a new framework to identify and assess progressive structural damage. The method relies solely on output measurements to establish the frequency response functions of a structure using cepstrum-based operational modal analysis. Two different damage indicative features are constructed using the established frequency response functions. The first damage feature takes the residual frequency response function, defined as the difference in frequency response function between evolving states of the structure, and then reduces its dimension using principle component analysis; while in the second damage indicator, a new feature based on the area under the residual frequency response function curve is proposed. The rationale behind this feature lies in the fact that damage often affects a number of modes of the system, that is, it affects the frequency response function over a wide range of frequencies; as a result, this quantity has higher sensitivity to any structural change by combining all contributions from different frequencies. The obtained feature vectors serve as inputs to a novel multi-stage neural network ensemble designed to assess the severity of damage in the structure. The proposed method is validated using extensive experimental data from a laboratory four-girder timber bridge structure subjected to gradually progressing damage at various locations with different severities. In total, 13 different states of the structure are considered, and it is demonstrated that the new damage feature outperforms the conventional principle component analysis–based feature. The contribution of the work is threefold: first, the application of cepstrum-based operational modal analysis in structural health monitoring is further validated, which has potential for real-life applications where only limited knowledge of the input is available; second, a new damage feature is proposed and its superior performance is demonstrated; and finally, the comprehensive test framework including extensive progressive damage cases validates the proposed technique.