Search

Summary A method for identifying the distribution of moving heavy vehicle loads is proposed for cable-stayed bridges based on a sparse l1 optimization technique. This method is inspired by the recently developed compressive sensing (CS) theory, which is a technique for obtaining sparse signal representations for underdetermined linear measurement equations. In this study, sparse l1 optimization is employed to localize the moving heavy vehicle loads of cable-stayed bridges through cable force measurements. First, a simplified equivalent load of vehicles on cable-stayed bridges is presented. Then, the relationship between the cable forces and the moving heavy vehicle loads is established based on the influence lines. With the hypothesis of a sparse distribution of vehicle loads on the bridge deck (which is practical for long-span bridges), moving heavy vehicle loads are identified by minimizing the ‘l2-norm'of the difference between the observed and simulated cable forces caused by the moving vehicles penalized by the ‘l1-norm’ of the moving heavy vehicle load vector. A numerical example of an actual cable-stayed bridge is employed to verify the proposed method. The robustness and accuracy of this identification approach (with measurement noise for multi-vehicle spatial localization) are validated. Copyright © 2015 John Wiley & Sons, Ltd.

SUMMARY Stay cables are critical components in bridges. However, stay cables suffer from severe fatigue damage. Therefore, a monitoring technique to obtain the time history of the tension in stay cables is important. Because the acceleration of stay cables is readily measurable, approaches to identify cable tension based on frequency analysis and monitored cable acceleration have been widely investigated and used in practice. However, this type of approach can only identify a time-invariant tension of a stay cable over a specified duration, not the time-varying tension. This paper proposes an approach to identify the time-varying tension of stay cables by monitoring cable accelerations. The tension variation in stay cables is caused by vehicles passing over the bridge. The real-time identification algorithm that determines the time-varying tension of stay cables is proposed using an extended Kalman filter based on both the transversal monitored acceleration at a single location on the cable and the monitored wind speed on the bridge, where the time-varying tension is a state variable that is identified. A stay cable from the Nanjing Yangtze River No. 3 Bridge was used for the numerical study. The time-varying tension of the stay cable can be identified when either a single vehicle or multiple vehicles pass over the bridge. The robustness of the proposed approach is also investigated through deviations in the initial tension, initial displacement, and velocity of the stay cable. An experiment was conducted on a scaled stay cable with time-varying tension excited by wind. The time-varying cable tension of the cable was identified by the proposed approach and compared with the real time-varying cable tension. The identification accuracy and robustness of the proposed approach is verified through the experiment and numerical study. Copyright © 2013 John Wiley & Sons, Ltd.

Summary This paper proposes an identification framework based on a restricted Boltzmann machine (RBM) for crack identification and extraction from images containing cracks and complicated background inside steel box girders of bridges. The original images that include fatigue crack and other background information are obtained by a consumer-grade camera inside the steel box girder. The original images are cut into a number of elements with small size as the input dataset, and a state representation vector is artificially labeled to every image element used for the crack identification. A deep learning model or network consisting of multiple processing RBM layers to learn the abstract features is constructed to match the input image elements with corresponding state representation vectors. Next, a three-layer RBM with 500; 500; and 2,000 hidden units is trained as the hidden layers in the deep learning network. A contrastive divergence learning algorithm is employed for training the deep network to update and obtain the optimal parameters (i.e., the biases and weights). The new input image elements labeled as crack are sorted out and assembled to form an output image. A deep network is modeled through the consumer-grade camera images containing cracks and complicated background information using the proposed approach. The accuracy and ability to identify cracks from new images with different resolutions using the trained deep network are validated. Furthermore, effects of element size on reconstruction error and identification accuracy are investigated. The results show that there exists optimal element size; that is, too small and too large element sizes both increase the reconstruction error and decrease the identification accuracy.