To quantitatively evaluate crack width of concrete structures surface, this paper presents a stereovision-based crack width detection method. Compared with the traditional visual inspection with single camera, this approach uses a pair of cameras to capture cracks images for recovering 3D coordinates of crack edge, and does not needs scale attached to concrete surface for converting measurement unit. A novel Canny-Zernike combination algorithm is utilized to obtain the image coordinates of crack edge in the left crack image, this combination algorithm can achieve 0.02 subpixel precision. The 3D coordinates of crack edge are acquired by projecting crack edge curve on concrete surface where cracks are located. The crack width is assessed by the minimum distance between two sides of crack edge. The detection tests are conducted on three concrete beams destroyed in static test, and the crack width of two inspection zones on each beam is acquired. Experimental results indicate that the stereovision-based crack width detection approach can accurately measure the crack width compared with the crack width gauge or the vernier calliper. This verifies the proposed method is applicable and useful for assessing the crack width of concrete surface.

Bridge displacement is the most basic evaluation index of the health status of a bridge structure. The existing measurement methods for bridge displacement basically fail to realize long-term and real-time dynamic monitoring of bridge structures, because of the low degree of automation and the insufficient precision, causing bottlenecks and restriction. To solve this problem, we proposed a bridge displacement monitoring system based on laser projection-sensing technology. First, the laser spot recognition method was studied. Second, the software for the displacement monitoring system was developed. Finally, a series of experiments using this system were conducted, and the results show that such a system has high measurement accuracy and speed. We aim to develop a low-cost, high-accuracy and long-term monitoring method for bridge displacement based on these preliminary efforts.

Small-scale models have been commonly utilized in testing of performance of active mass driver (AMD) control systems. The utmost reason is that physical testing of AMD system at full scale is usually too expensive to afford, as well as hard to implement on site. With reference to the real-time hybrid simulation technology, a real-time AMD subsystem testing method is proposed in this paper. In this method, the entire system is composed of AMD as physical subsystem and target structure as numerical subsystem. The physical test is conducted on AMD subsystem, whereas the numerical simulation is carried out on the structure subsystem. Meanwhile, the real-time data are being communicated between these two subsystems. This method is then applied to the performance validation of a novel AMD control system, which is driven by a linear motor. In the test, a benchmark three-storey frame structure is employed as the numerical subsystem, and earthquake excitations are used as the external input. On the basis of a series of tests, both the time history and the statistical criteria show that the results of AMD subsystem and structure subsystem obtained in the real-time AMD subsystem test agree well with the simulation results. Furthermore, all the test results show good repeatability. Therefore, the feasibility and reliability of the proposed real-time AMD subsystem testing approach for performance validation of AMD subsystem has been demonstrated. Such kind of experimental method is efficient in terms of reducing cost associated with performance validation of large-scale active control systems prior to the implementation. Copyright © 2013 John Wiley & Sons, Ltd.

Structural damage detections based on the changes of dynamic properties are a major concern for structural health monitoring. In this paper, efforts are made to extend the flexibility-based damage localization methods, especially the damage locating vector (DLV) method, to the case of ambient vibration with incomplete measured degree of freedom, where flexibility matrices are not available. First, the method to assemble a proportional flexibility matrix (PFM) with arbitrarily scaled modal shapes of full measured degrees of freedom (DOF) is introduced. The PFM is within a scalar multiplier to the real flexibility matrix, and the multiplier is shown to be the first modal mass theoretically. In the case of incomplete measured degrees, it is proved that the FEM model with full measured DOFs has the same modal masses and modal stiffnesses as the condensed model with partially measured DOFs as the retained degrees. Based on this deduction, the PFM at sensor locations is achieved with arbitrarily scaled test modes at partially measured DOFs. Assuming that the modal masses do not change significantly before and after being damaged, the PFMs for pre- and post-damage structures are made comparable, and the DLV method is implemented with the proposed PFMs at sensor locations. Finally an example of multi-damage sites localization for a 14 bays planar truss is given. Five damaged members in this structure are successfully identified by the proposed approach with only outputs and measurements at partial DOFs. Copyright © 2005 John Wiley & Sons, Ltd.