Structural damage detection based on convolutional neural networks and population of bridges. (October 2022)

• Record Type: Journal Article
• Authors: Teng, Shuai; Chen, Xuedi; Chen, Gongfa; Cheng, Li; Bassir, David
• Is Part Of: Measurement. Volume 202(2022)
• Publication Date: 2022-10
• Subjects: Structural damage detection -- Population of bridges -- Convolutional neural network -- Bridge structure -- Vibration signal; SHM Structural Health Monitoring -- CNN Convolutional Neural Network -- SDD Structural Damage Detection -- UT Ultrasonic Testing -- AE Acoustic Emissions -- LT Laser Testing -- ML Machine Learning -- FE Finite Element -- ACC Acceleration -- DS Displacement -- MS Mode Shape -- MSE Modal Strain Energy -- SM Strain Mode -- FC Fully Connection -- Conv Convolution -- FSB Flat Steel Bar; Weights and measures -- Periodicals; Measurement -- Periodicals; Measurement; Weights and measures; Periodicals; 530.8
• DOI: 10.1016/j.measurement.2022.111747
• Languages: English
• ISSNs: 0263-2241

Abstract

Highlights: The damage features of the bridge population were extracted by the CNN; and the CNN was applied to damage detection of the new randomly-created models. The detection accuracy of multiple damage indicators was also compared. The robustness of the proposed method was confirmed by the noise interference experiment and k -fold cross validation. Abstract: The CNN-based detection methods have been widely used in the field of structural health monitoring (SHM), however, they can only be used for individual structures under certain conditions; for the structures in-service, damage detection will be affected by a variety of external factors (unknown/uncertain load and geometric dimension, etc.). Therefore, in order to improve the applicability of the CNNs, their compatibility and robustness need to be thoroughly investigated. In this paper, a large number of random models were produced to establish a population of bridge structures, the damage features of the population were extracted by the CNN; subsequently, the CNN was applied to damage detection of the new randomly-created models. The results show that the best detection results (99.4% accuracy) can be obtained by using the acceleration signals as the CNN input. This demonstrates that the proposed method will expand the detection ability of the CNN beyond an individual structure.