Structural damage detection of steel bridge girder using artificial neural networks and finite element models

Abstract

Damage in structures often leads to failure. Thus it is very important to monitor structures for the occurrence of damage. When damage happens in a structure the consequence is a change in its modal parameters such as natural frequencies and mode shapes. Artificial Neural Networks (ANNs) are inspired by human biological neurons and have been applied for damage identification with varied success. Natural frequencies of a structure have a strong effect on damage and are applied as effective input parameters used to train the ANN in this study. The applicability of ANNs as a powerful tool for predicting the severity of damage in a model steel girder bridge is examined in this study. The data required for the ANNs which are in the form of natural frequencies were obtained from numerical modal analysis. By incorporating the training data, ANNs are capable of producing outputs in terms of damage severity using the first five natural frequencies. It has been demonstrated that an ANN trained only with natural frequency data can determine the severity of damage with a 6.8% error. The results shows that ANNs trained with numerically obtained samples have a strong potential for structural damage identification.

Residual Strength Prediction of Aluminum Panels with Multiple Site Damage Using Artificial Neural Networks

“…Since its introduction in the 1960s, ANNs continued to provide a powerful framework for modeling nonlinear systems, and they were used in a wide variety of engineering applications, including automatic control [ 24 ], solar energy systems [ 25 ], traffic and transportation [ 26 ], image processing [ 27 ], optimization of structures [ 28 ], materials science and engineering [ 29 , 30 , 31 , 32 ], manufacturing [ 33 ], fracture mechanics, and fault detection [ 34 , 35 , 36 , 37 , 38 , 39 , 40 , 41 , 42 , 43 , 44 , 45 , 46 , 47 , 48 , 49 , 50 , 51 ]. In fracture mechanics, ANNs were mostly used in applications concerned with crack propagation, fatigue life, and failure mode prediction [ 34 ].…”

“…Bayesian regularization (BR), Levenberg–Marquardt (LM), and Scaled Conjugate Gradient (SCG) algorithms demonstrated high effectiveness in predicting the failure load of adhesively bonded composites. Hakim and Razak [ 36 ] proposed ANN to predict the damage severity of a steel girder bridge. The inputs of the network were the first five natural frequencies, while the output parameter was a damage index.…”

Railway bridge structural health monitoring and fault detection: State-of-the-art methods and future challenges

“…Although the method, which was tested on a laboratory model of a bridge, showed good performance in the localization of the failure using the strain as the input to the ANN, only the first two levels of the four requirements of the fault detection process were accomplished, that is, the magnitude of the failure and the future behaviour of the bridge were not assessed. Hakim and Abdul Razak 107 developed a BPNN in order to identify failure and to assess severity of a steel girder bridge. Results showed that the trained ANN was able to identify with high accuracy (higher than 90%) the failure severity of the test patterns.…”

“…This procedure is one of the most critical aspects of the use the ANN, and it is generally addressed using a trial-and-error procedure. 2,105–109 In order to point out a methodological criterion to choose the optimal ANN structure, a Bayesian process can be adopted by considering that an increasing number of the hidden layers leads to higher accuracy of data fitting, but poor predictions for new failure cases may be achieved due to an over-parameterization of the problem. 117 Indeed, if the ANN is over-parameterized, that is, the number of hidden nodes is too high, the ANN is optimal to mimic the training set, but it is unable to manage new and unknown patterns.…”

Localizing and quantifying structural damage by means of a beetle swarm optimization algorithm

“…The changes in modal parameters and their combinations are widely utilized to construct an objective function (Frigui et al, 2018; Meruane and Heylen, 2011; Pandey and Biswas, 1994; Pandey et al, 1991; Perera et al, 2009; Shabbir and Omenzetter, 2016; Villalba and Laier, 2012). Among them, the natural frequencies are easier to be measured and have high sensitivity to damage (Hakim and Razak, 2013); besides, their measurement errors are negligible compared to that of mode shapes (Perera et al, 2009). Therefore, these frequencies are widely used for damage identification at first.…”