Young-Jin Cha

Deep Learning-Based Crack Damage Detection Using Convolutional Neural Networks

A number of image processing techniques IPTs have been implemented for detecting civil infrastructure defects to partially replace human-conducted onsite inspections. These IPTs are primarily used to manipulate images to extract defect features, such as cracks in concrete and steel surfaces. However, the extensively varying real-world situations e.g., lighting and shadow changes can lead to challenges to the wide adoption of IPTs. To overcome these challenges, this article proposes a vision-based method using a deep architecture of convolutional neural networks CNNs for detecting concrete cracks without calculating the defect features. As CNNs are capable of learning image features automatically, the proposed method works without the conjugation of IPTs for extracting features. The designed CNN is trained on 40 K images of 256 × 256 pixel resolutions and, consequently, records with about 98% accuracy. The trained CNN is combined with a sliding window technique to scan any image size larger than 256 × 256 pixel resolutions. The robustness and adaptability of the proposed approach are tested on 55 images of 5,888 × 3,584 pixel resolutions taken from a different structure which is not used for training and validation processes under various conditions e.g., strong light spot, shadows, and very thin cracks. Comparative studies are conducted to examine the performance of the proposed CNN using traditional Canny and Sobel edge detection methods. The results show that the proposed method shows quite better performances and can indeed find concrete cracks in realistic situations.

Structural Damage Detection Using Modal Strain Energy and Hybrid Multiobjective Optimization

Modal strain energy (MSE) is a sensitive physical property that can be utilized as a damage index in structural health monitoring. Inverse problem solving-based approaches using single-objective optimization algorithms are also a promising damage identification method. However, the research into the integration of these methods is currently limited; only partial success in the detection of structural damage with high errors has been reported. The majority of previous research was focused on detecting damage in simply supported beams or plain structures. In this study, a novel damage detection approach using hybrid multiobjective optimization algorithms based on MSE is proposed to detect damages in various three-dimensional (3-D) steel structures. Minor damages have little effect on the difference of the modal properties of the structure, and thus such damages with multiple locations in a structure are difficult to detect using traditional damage detection methods based on modal properties. Various minor damage scenarios are created for the 3-D structures to investigate the newly proposed multiobjective approach. The proposed hybrid multiobjective genetic algorithm detects the exact locations and extents of the induced minor damages in the structure. Even though it uses incomplete mode shapes, which do not have any measured information at the damaged element, the proposed approach detects damage well. The robustness of the proposed method is investigated by adding 5% Gaussian random white noise as a noise effect to mode shapes, which are used in the calculation ofMSE.

Multi-objective genetic algorithms for cost-effective distributions of actuators and sensors in large structures

This paper proposes a multi-objective genetic algorithm (MOGA) for optimal placements of control devices and sensors in seismically excited civil structures through the integration of an implicit redundant representation genetic algorithm with a strength Pareto evolutionary algorithm 2. Not only are the total number and locations of control devices and sensors optimized, but dynamic responses of structures are also minimized as objective functions in the multi-objective formulation, i.e., both cost and seismic response control performance are simultaneously considered in structural control system design. The linear quadratic Gaussian control algorithm, hydraulic actuators and accelerometers are used for synthesis of active structural control systems on large civil structures. Three and twenty-story benchmark building structures are considered to demonstrate the performance of the proposed MOGA. It is shown that the proposed algorithm is effective in developing optimal Pareto front curves for optimal placement of actuators and sensors in seismically excited large buildings such that the performance on dynamic responses is also satisfied.

Comparative Studies of Semiactive Control Strategies for MR Dampers: Pure Simulation and Real-Time Hybrid Tests

AbstractThis paper presents comparisons of the performances of three semiactive control algorithms for use with multiple magnetorheological (MR) dampers. The three controllers are (1) the clipped-optimal controller, (2) the decentralized output feedback polynomial controller, and (3) the simple passive controller. These controllers use different types of inputs to calculate control signals for the MR dampers, based on different control mechanisms. To investigate the advantages of each controller, a three-degree-of-freedom steel moment-resisting frame, designed using a performance-based design methodology, was developed. The performance was investigated by using four different earthquakes utilized during the design of the building frame. Comparisons of the controllers’ performance were carried out in terms of reduction in the maximum interstory drifts, displacements, absolute accelerations, and control forces. Real-time hybrid tests were carried out to validate these comparisons.

Multi-objective optimization for actuator and sensor layouts of actively controlled 3D buildings

This paper investigates the multi-objective optimization of active control systems for vibration control of three-dimensional (3D) high-rise buildings under a variety of earthquake excitations. To this end, a novel multi-objective genetic algorithm is developed through the integration of the best features of a non-dominated sorting II (NS2) genetic algorithm (GA) and an implicit redundant representation (IRR) GA. The proposed NS2-IRR GA finds not only minimum distributions of both actuators and sensors within structures, but also minimum dynamic responses of 3D structures. Linear quadratic Gaussian controllers, hydraulic actuators and accelerometers are used for implementation of active control systems within the 3D buildings. To demonstrate the effectiveness of the proposed NS2-IRR GA, two 3D building models are investigated using finite element methods, including low- and high-rise buildings. It is shown that the proposed NS2-IRR GA is effective in finding not only optimal locations and numbers of both actuators and sensors in 3D buildings, but also minimum responses of the 3D buildings. The simulation also shows that the control performances of the proposed approach significantly enhance those of the engineering judgment oriented benchmark layout, which is validated by comparisons of each performance using the same number of actuators.

Large-Scale Real-Time Hybrid Simulation for Evaluation of Advanced Damping System Performance

AbstractAs magnetorheological (MR) control devices increase in scale for use in real-world civil engineering applications, sophisticated modeling and control techniques may be needed to exploit their unique characteristics. Here, a control algorithm that utilizes overdriving and backdriving current control to increase the efficacy of the control device is experimentally verified and evaluated at large scale. Real-time hybrid simulation (RTHS) is conducted to perform the verification experiments using the nees@Lehigh facility. The physical substructure of the RTHS is a 10-m tall planar steel frame equipped with a large-scale MR damper. Through RTHS, the test configuration is used to represent two code-compliant structures, and is evaluated under seismic excitation. The results from numerical simulation and RTHS are compared to verify the RTHS methodology. The global responses of the full system are used to assess the performance of each control algorithm. In each case, the reduction in peak and root mean s...

Wavelet-neuro-fuzzy control of hybrid building-active tuned mass damper system under seismic excitations

This paper proposes a wavelet-based fuzzy neuro control algorithm for the hazard mitigation of seismically excited buildings equipped with a hybrid control system. The wavelet-based fuzzy neuro controller is developed through the integration of discrete wavelet transform, artificial neural network, and a Takagi-Sugeno fuzzy controller. The hybrid control system is an integrated model of an actuator, a tuned mass damper, and viscous liquid dampers: an active tuned mass damper (ATMD) is located on the top floor of the structure and viscous liquid dampers are located on each floor. To demonstrate the effectiveness of the proposed wavelet-based adaptive neuro-fuzzy inference system (WANFIS) controller, an eight-story building employing passive viscous liquid dampers as well as an ATMD is investigated. A variety of earthquakes such as an artificial earthquake, the 1940 El-Centro, Kobe, Northridge, and Hachinohe earthquakes are used as disturbance signals. It is demonstrated that the WANFIS controller is effective in reducing the structural responses of the hybrid structure system subjected to a variety of disturbances.

Structural Modal Identification Through High Speed Camera Video: Motion Magnification

Video cameras offer the unique capability of collecting high density spatial data from a distant scene of interest. They could be employed as remote monitoring or inspection sensors because of their commonplace use, simplicity, and relatively low cost. The difficulty is in interpreting the video data into a usable format that is familiar to engineers such as displacement. A methodology called motion magnification, developed for visualizing exaggerated versions of small displacements, is extended to modal identification in structures. Experiments in a laboratory setting on a cantilever beam were performed to verify the method against accelerometer and laser vibrometer measurements. Motion magnification is used for modal analysis of cantilever beams to visualize mode shapes and calculate mode shape curvature as a basis for damage detection. Suggestions for applications of this methodology and challenges in real-world implementations are given.

Autonomous Structural Visual Inspection Using Region-Based Deep Learning for Detecting Multiple Damage Types

Computer vision-based techniques were developed to overcome the limitations of visual inspection by trained human resources and to detect structural damage in images remotely, but most methods detect only specific types of damage, such as concrete or steel cracks. To provide quasi real-time simultaneous detection of multiple types of damages, a Faster Region-based Convolutional Neural Network (Faster R-CNN)-based structural visual inspection method is proposed. To realize this, a database including 2,366 images (with 500 × 375 pixels) labeled for five types of damages—concrete crack, steel corrosion with two levels (medium and high), bolt corrosion, and steel delamination—is developed. Then, the architecture of the Faster R-CNN is modified, trained, validated, and tested using this database. Results show 90.6%, 83.4%, 82.1%, 98.1%, and 84.7% average precision (AP) ratings for the five damage types, respectively, with a mean AP of 87.8%. The robustness of the trained Faster R-CNN is evaluated and demonstrated using 11 new 6,000 × 4,000-pixel images taken of different structures. Its performance is also compared to that of the traditional CNN-based method. Considering that the proposed method provides a remarkably fast test speed (0.03 seconds per image with 500 × 375 resolution), a framework for quasi real-time damage detection on video using the trained networks is developed.

Active control of highway bridges subject to a variety of earthquake loads

In this paper, a wavelet-filtered genetic-neuro-fuzzy (WGNF) control system design framework for response control of a highway bridge under various earthquake loads is discussed. The WGNF controller is developed by combining fuzzy logic, discrete wavelet transform, genetic algorithms, and neural networks for use as a control algorithm. To evaluate the performance of the WGNF algorithm, it is tested on a highway bridge equipped with hydraulic actuators. It controls the actuators installed on the abutments of the highway bridge structure. Various earthquakes used as input signals include an artificial earthquake, the El-Centro, Kobe, North Palm Springs, Turkey Bolu, Chi-Chi, and Northridge earthquakes. It is proved that the WGNF control system is effective in mitigating the vibration of the highway bridge under a variety of seismic excitation.