Abstract
A customized YOLOv8-seg architecture is hired in this study to automatically detect and segment post-earthquake damage, such as cracks, spalling, reinforcement exposure, crushing, buckling, and structural failure, that appears on bridge piers tested using slow and fast cyclic tests, shaking table tests, and real-time hybrid simulations. Using a hybrid loss function, the YOLOv8-seg model processes 32×32 and 256×256-pixel image patches, extracted from 124 large RGB images, for cracks and other seismic damage categories, respectively. Training is conducted on the image patches and their corresponding labeled annotations, distinguishing between seismic damage and background (non-damage) pixels. The model is trained with a batch size of 16, utilizing the Adamax optimizer, an exponential learning rate scheduler, and weight decay techniques to improve training stability and performance. The results demonstrate that the generated mask patches closely resemble the actual damage patterns.
Address
(1) Omid Yazdanpanah:
Hybrid Structural Testing Center (Hystec), Myongji University, Republic of Korea;
(2) Ensieh Ali Bakhshi:
Industry & Academia Cooperation Foundation, Myongji University, Republic of Korea;
(3) Minwoo Chang:
Department of Civil and Environmental Engineering, Myongji University, Republic of Korea;
Abstract
Aiming at the low accuracy of fault identification caused by insufficient fault feature extraction in vibration signals of rolling bearings, a fault diagnosis method based on whale algorithm to optimize variational modal decomposition parameters (WOA-VMD) for feature extraction and convolution neural network coupled with support vector machine (CNN-SVM) is proposed. Firstly, the parameters of VMD are optimized by WOA algorithm, and then some intrinsic modal components (IMF) are obtained by decomposing the fault signal by the VMD method. Then the IMF components are screened by correlation coefficient method, and the sample envelope entropy is further extracted as the feature vector. Finally, CNN-SVM classifier is used as a fault identification method to identify the faults of rolling bearings. The experimental results show that the WOAVMD feature extraction method can accurately extract the fault information of rolling bearing vibration signals, and CNN-SVM classifier can effectively identify the fault features in bearing vibration signals. Compared with SVM and PSO-SVM classification methods, the proposed method can improve the fault recognition rate, and the accuracy rate can be improved to 99.6%.
Key Words
convolutional neural network; rolling bearing failure; sample envelope entropy; support vector machine; variational modal decomposition; whale optimization algorithm
Address
Dalian Scientific Test and Control Technology Institute, Dalian 116013, China.
Abstract
Construction sites present diverse and evolving visual conditions that challenge the generalizability of pre-trained object detection models. This study proposes a parameter-efficient fine-tuning approach based on Low-Rank Adaptation to enable adaptive object detection tailored to site-specific conditions. A general model was trained on a large-scale dataset and fine-tuned using both the proposed method and full fine-tuning across three real-world construction projects. Despite utilizing only 12% of the trainable parameters, the proposed approach achieved comparable or superior detection accuracy with 10% reduced training time and 30% lower GPU memory consumption. These results highlight its effectiveness in adapting object detection models to site-specific conditions under resource constraints. Furthermore, the approach can be extended with semisupervised learning to support scalable adaptation in construction environments.
Key Words
adaptive object detection; computer vision; construction safety Low-Rank Adaptation (LoRA); parameterefficient fine-tuning
Address
Department of Civil and Environmental Engineering, Korea Advanced Institute for Science and Technology, 291 Daehak-ro, Yuseong-gu, Daejeon 305-701, Republic of Korea.
Abstract
This research introduces a smart control paradigm for the stochastic vibro-acoustic suppression of functionally graded piezoelectric (FGP) concrete plates, which are developed for the smart concrete structures and systems domain. The proposed method integrates negative capacitance piezoelectric shunt damping (NCPSD), the Carrera Unified Formulation (CUF), and a smart hybrid optimization mechanism comprising deep neural networks and Genetic algorithms (DNN-GA). A detailed multi-layer modeling framework is created through CUF, which efficiently depicts the complex electromechanical responses such as shear deformation, geometric nonlinearity, and spatial grading in the piezoelectric properties. A smart passiveactive damping interface is realized by embedding piezoelectric sensor-actuator pairs connected to a negative capacitance-resistive-inductive (NC-RL) shunt circuit. This setup significantly boosts dynamic adaptability and provides broad bandwidth attenuation. The DNN-GA architecture controls the parameters by tuning them adaptively according to the stochastic excitations and thus navigating through the complex nonlinear response space of the FGP concrete system effectively. The genetic algorithm proceeds rapidly through the zones of the optimal solutions while the deep neural network ensures real-time prediction and adaptation amid the parametric uncertainties. There was a considerable reduction in structural vibration and radiation of acoustic energy particularly in the mid-to-high frequency range, as simulation results indicated. This research supports the possibility of using smart damping solutions for smart concrete structures and systems abroad, especially in the aerospace and automotive industries where the ability to reduce noise and vibrations adaptively is needed the most.
Key Words
DNN-GA algorithm; intelligent structural dynamics; functionally graded piezoelectric concrete plates; negative capacitance shunt damping; smart concrete structures and systems; stochastic vibro-acoustic control
Address
(1) Bin Wang:
School of Civil Engineering and Transportation Engineering, Yellow River Conservancy Technical University, Kaifeng 475004, Henan, China;
(2) Bo Han:
College of Science, Xi'an University of Architecture and Technology, Xi'an 710311, Shanxi Province, China;
(3) Mehran Safarpour:
Faculty of Engineering, Department of Mechanics, Tarbiat Modares University, Tehran, Iran;
(4) Murat Yaylacı:
Department of Civil Engineering, Recep Tayyip Erdogan University, 53100, Rize, Turkey;
(5) Murat Yaylacı:
Turgut Kiran Maritime Faculty, Recep Tayyip Erdogan University, 53900, Rize, Turkey.
