This study proposes a new algorithm for damage detection in structures. The algorithm employs an energy-based method to capture linear and nonlinear effects of damage on structural response. For more accurate detection the proposed algorithm combines multiple damage sensitive features through a distance-based method by using Mahalanobis distance. Hypothesis testing is employed as the statistical data analysis technique for uncertainty quantification associated with damage detection. Both the distance-based and the data analysis methods have been chosen to deal with small size data sets. Finally, the efficacy and robustness of the algorithm is experimentally validated by testing a steel laboratory prototype and the results show that the proposed method can effectively detect and localize the defects. KeywordsEnergy method, hypothesis testing, marginal Hilbert spectrum, normalized cumulative energy distribution, Mahalanobis distance, white noise excitation 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 IntroductionVibration based structural health monitoring (SHM) is a widely used method for monitoring large scale, complex structures. Aging of infrastructures, higher operational demands, and variety of environmental effects on structural systems are the main reasons that attract more attention to this field in recent years.The algorithms for vibration-based SHM are either model-based or data-based. Both methods compare the response of the system with a baseline. In model-based approach, the baseline is provided using numerical models. Thus, this method is helpful for systems for which the model already exists and in cases where it is justifiable to build a sufficiently accurate structural model [1][2][3]. The data-based approach brings more flexibility to the damage detection scheme since it only uses the sensor data without having to deal with the complications of creating a model. The initial phase in both of these methodologies is feature extraction. In this phase certain damage sensitive features, called damage index (DI), are extracted from the structure's response, either empirically obtained or numerically simulated, to measure its discrepancies from the response in the intact state. Previous studies show that the features which capture nonlinearities in the structural response are generally more sensitive to damage, less sensitive to environmental conditions, and hence, more reliable for the purpose of damage detection compared to the DIs that capture linear phenomena such as modal properties [4][5][6][7].Note that the source of nonlinearities can be material, geometry, or nonlinear dynamics phenomenon such as dispersion, mode mixing, and damping. The fractal dimension of the attractor of time-series is the basis for defining DIs in [8][9][10][11]. Fractal analysis of residual crack patterns in reinforced concrete struct...

Through advances in sensor technology and development of camera-based measurement techniques, it has become affordable to obtain high spatial resolution data from structures. Although measured datasets become more informative by increasing the number of sensors, the spatial dependencies between sensor data are increased at the same time. Therefore, appropriate data analysis techniques are needed to handle the inference problem in presence of these dependencies. In this paper, we propose a novel approach that uses graphical models (GM) for considering the spatial dependencies between sensor measurements in dense sensor networks or arrays to improve damage localization accuracy in structural health monitoring (SHM) application. Because there are always unobserved

A hardware system has been developed to obtain data from multiple MEMS triaxial accelerometers for structural health monitoring research in a laboratory setting. The system can be easily configured for a single accelerometer or up to as many as 120 triaxial accelerometers with 24-bit data sampled up to a 2 kHz rate simultaneously. The system is modular where each Aggregator Module consists of a circuit board that supports up to eight accelerometers.Up to 15 modules can be synchronized to within 0.1 µs for simultaneous data acquisition and timestamping by connecting a central clock module that distributes a 10 MHz clock to each Aggregator Module. This represents (4 channels per ADC) x (8 ADCs) x (15 Aggregator Modules) = 480 channels that are all simultaneously sampled and synchronized down to 0.1 µs. The data from each Aggregator Module is transmitted to a single host computer or multiple host computers using serial communication. The advantages of this custom system over a wireless solution are its very tight synchronization, high sample rate, high resolution and ability to process large quantities of data. COTS wired solutions do offer these features; however, to accommodate 480 sensor channels, numerous bulky and expensive pieces of equipment are needed. We decided that a custom design allowed us the greatest performance in a scalable, lower power, portable and affordable package.The original intended deployment of this custom system was in a laboratory setting, i.e. distances of one to three meters. However, we are extending the utility of the system to real world civil structures such as a parking garage or a pedestrian bridge with characteristic length scales of 10's of meters. The primary constraint on the physical size of the deployment is the cable length from the central clock module to the various Aggregator Modules which directly impacts the signal integrity of the central clock. Laboratory testing has shown the clock module can drive 15.25m (50ft) of RG58 coaxial cable without signal degradation. Using this limit we will deploy an accelerometer network on a real world civil structure and report the data from that network.We have also developed a strategy for surpassing the 15.25m limit by daisy chaining modules and using the clock/synchronization module's FPGA to regenerate the clock for the next Aggregator Module in the daisy chain. Although the clock regeneration at each module will create a clock delay, this delay is deterministic and can be accounted for when postprocessing the data. We will report on laboratory test results showing the clock signal integrity over distances greater than 15.25 meters.Our installation will demonstrate the scalability and portability of our custom, highly synchronous, simultaneously sampled data collection system in a real world deployment. A wireless system would not be capable of maintaining the tight synchronization or handling the large amounts of data, and a COTS wired solution would be less portable and more costly.

In this study, a new damage detection algorithm for specific types of damages such as breathing cracks, which are called “active discontinuities” in this paper, is proposed. The algorithm is based on the nonlinear behavior of this class of damages and hence, is more precise and sensitive to damage compared to other common linear methods. The active discontinuities can be regarded as additional degrees of freedom (DOFs) which need energy to be excited. Because the input energy of both the intact and the damaged structures is finite, the energy content of vibrating modes will be changed due to damage. Thus, the properties of distribution of energy between vibrating modes can be used as indices for detecting damage. An essential detectability condition using this concept is decomposing a signal such that no spurious mode imposed to its expansion. In order to satisfy this condition, Empirical Mode decomposition (EMD) is used to extract the vibrating modes since all nonlinearities in a signal are preserved while no spurious mode or assumption of stationarity is imposed on the problem. Prevention of mode mixing, which is an important drawback of EMD, is another necessary condition for robustness of the algorithm. A solution is proposed in this paper to satisfy this condition in which special constraints are imposed on the normal procedure of EMD. Then, the fourth central moment, kurtosis, is used to compare the distribution of energy between the modified vibrating modes. The algorithm is verified through experimental testing of a simple steel cantilever structure under various damage scenarios. Results demonstrate the efficacy of the method for detecting discontinuities in a real structure.

A method for localization and severity assessment of structural damages is proposed. The algorithm works based on nonlinear behavior of certain type of damages such as breathing cracks which are called active discontinuities in this paper. Generally, nonlinear features are more sensitive to such damages although their extraction is sometimes controversial. A major controversy is the imposition of spurious modes on the expansion of the signal which needs to be addressed for an effective application and robustness of the method. The energy content of Intrinsic Mode Functions (IMFs), which are the resultants of Empirical Mode Decomposition (EMD), and also the shape of energy distribution between these modes before and after damage, are used for localization and severity assessment of the damages. By using EMD, we preserve the nonlinear aspects of the signal while avoiding imposition of spurious harmonics on its expansion without any assumption of stationarity. The developed algorithms are used to localize and assess the damage in a steel cantilever beam. The results show that the method can be used effectively for detecting active structural discontinuities due to damage.