Showing papers by "Charles R. Farrar published in 2018"

Reference-free detection of minute, non-visible, damage using full-field, high-resolution mode shapes output-only identified from digital videos of structures:

Abstract: Detecting damage in structures based on the change in their dynamics or modal parameters (modal frequencies and mode shapes) has been extensively studied for three decades. The success of such a global, passive, vibration-based method in field applications, however, has long been hindered by the bottleneck of low spatial resolution vibration sensor measurements. The primary reason is that damage typically initiates and develops in local regions that need to be captured and characterized by very high spatial resolution vibration measurements and modal parameters (mode shapes), which are extremely difficult to obtain using traditional vibration measurement techniques. For example, accelerometers and strain-gauge sensors are typically placed at a limited number of discrete locations, providing low spatial resolution vibration measurements. Laser vibrometers provide high-resolution measurements, but are expensive and make sequential measurements that are time- and labor-consuming. Recently, digital video cameras—which are relatively low cost, agile, and able to provide high spatial resolution, simultaneous, pixel measurements—have emerged as a promising tool to achieve full-field, high spatial resolution vibration measurements. Combined with advanced vision processing and unsupervised machine algorithms, a new method has recently been developed to blindly and efficiently extract the full-field, high-resolution, dynamic parameters from the video measurements of an operating, output-only structure. This work studies the feasibility of performing damage detection using the full-field, very high spatial resolution mode shape (of the fundamental mode) blindly extracted from the video of the operating (output-only) structure without any knowledge of reference (healthy) structural information. A spatial fractal dimension analysis is applied on the full-field mode shape of the damaged structure to detect damage-induced irregularity. Additionally, the equivalence between the fractal dimension and the squared curvature (modal strain energy) of the mode shape curve, when of high spatial resolution, is mathematically derived. Laboratory experiments are conducted on bench-scale structures, including a building structure and a cantilever beam, to validate the approach. The results illustrate that using the full-field, very high-resolution mode shape enables detection of minute, non-visible, damage in a global, completely passive sensing manner, which was previously not possible to achieve.

Remote railroad bridge structural tap testing using aerial robots

Abstract: Infrastructure spending is a major component of railroad annual budgets and therefore must be prioritized to ensure both safety and line capacity at all times. Current inspection priorities for railroads include early detection of structural deterioration, including concrete and timber, caused by the recent increase of car load capacities. Railroad bridge inspectors conduct tap testing to detect the deterioration of concrete and timber acoustically. However, measuring railroad bridge condition in the field is a challenging task and, in general, is solely based on the inspector’s experience. The results of this research describe the development, validation, and testing of a remote tap testing device that can be deployed by an aerial robot. The new tap testing device can remotely impact the surface, record the sounds of those impacts, and post-process the data to perform replacement prioritization. Tapping acoustics were analyzed both in the time and frequency domains. Principal component analysis of the data enabled the clustering of the different sets of data collected from different concrete and timber surfaces. The results quantify for the first time structural tap testing data collected with a tap testing mechanism built for deployment by an aerial robot. The goal of this work is to enable cost-effective, safer and sustainable upgrade prioritization of railroad bridge inventories.

Efficient Full-Field Vibration Measurements and Operational Modal Analysis Using Neuromorphic Event-Based Imaging

Abstract: Traditional vibration measurement typically requires physically attached sensors, such as accelerometers and strain gauges. However, these discrete-point sensors provide only low spatial re...

Spatiotemporal video-domain high-fidelity simulation and realistic visualization of full‐field dynamic responses of structures by a combination of high-spatial-resolution modal model and video motion manipulations

Abstract: Structures with complex geometries, material properties, and boundary conditions exhibit spatially local dynamic behaviors. A high‐spatial‐resolution model of the structure is thus required for high‐fidelity analysis, assessment, and prediction of the dynamic phenomena of the structure. The traditional approach is to build a highly refined finite element computer model for simulating and analyzing the structural dynamic phenomena based on detailed knowledge and explicit modeling of the structural physics such as geometries, materials properties, and boundary conditions. These physics information of the structure may not be available or accurately modeled in many cases, however. In addition, the simulation on the high‐spatial‐resolution structural model, with a massive number of degrees of freedom and system parameters, is computationally demanding. This study, on a proof‐of‐principle basis, proposes a novel alternative approach for spatiotemporal video‐domain high‐fidelity simulation and realistic visualization of full‐field structural dynamics by an innovative combination of the fundamentals of structural dynamic modeling and the advanced video motion manipulation techniques. Specifically, a low‐modal‐dimensional yet high‐spatial (pixel)‐resolution (as many spatial points as the pixel number on the structure in the video frame) modal model is established in the spatiotemporal video domain with full‐field modal parameters first estimated from line‐of‐sight video measurements of the operating structure. Then in order to simulate new dynamic response of the structure subject to a new force, the force is projected onto each modal domain, and the modal response is computed by solving each individual single‐degree‐of‐freedom system in the modal domain. The simulated modal responses are then synthesized by the full‐field mode shapes using modal superposition to obtain the simulated full‐field structural dynamic response. Finally, the simulated structural dynamic response is embedded into the original video, replacing the original motion of the video, thus generating a new photo‐realistic, physically accurate video that enables a realistic, high‐fidelity visualization/animation of the simulated full‐field vibration of the structure. Laboratory experiments are conducted to validate the proposed method, and the error sources and limitations in practical implementations are also discussed. Compared with high‐fidelity finite element computer model simulations of structural dynamics, the video‐based simulation method removes the need to explicitly model the structure's physics. In addition, the photo‐realistic, physically accurate simulated video provides a realistic visualization/animation of the full‐field structural dynamic response, which was not traditionally available. These features of the proposed method should enable a new alternative to the traditional computer‐aided finite element model simulation for high‐fidelity simulating and realistically visualizing full‐field structural dynamics in a relatively efficient and user‐friendly manner.

System and method for automated extraction of high resolution structural dynamics from video

Abstract: A method for extracting vibrational modes of a structure includes: receiving a plurality of video frames, each of the video frames including a plurality of pixels; decomposing each of the video frames on a plurality of spatial scales in accordance with complex steerable pyramid filters to obtain a filter response for each of the spatial scales; computing a plurality of local phases of the pixels of each frame; removing a temporal mean from each frame to obtain a plurality of factored vibration motion functions; performing principal component analysis on the factored vibration motion functions to obtain principal components; blind source separating the principal components to compute a plurality of modal coordinates; computing frequency and damping ratios in accordance with the modal coordinates; and outputting the computed frequency and damping ratios.