Abstract
This study presents a novel approach to enhancing the seismic performance of tuned mass damper (TMD) systems in reinforced concrete (RC) structures through the implementation of a Smart Lyapunov Linear Matrix Inequality (LMI) criterion, optimized via deep reinforcement learning (DRL) algorithms. Traditional methods for TMD design often rely on heuristic or empirical approaches, which may not adequately address the complexities of dynamic interactions in RC structures under varying seismic loads. By leveraging the capabilities of DRL, this research develops a framework that dynamically adjusts TMD parameters in real-time, ensuring optimal performance across a range of seismic scenarios. The proposed Smart Lyapunov LMI criterion provides a robust mathematical foundation for stability and performance assessment, allowing for the systematic evaluation of TMD effectiveness in mitigating structural vibrations. Through extensive numerical simulations, the integration of DRL algorithms demonstrates significant improvements in the adaptability and efficiency of TMD systems, outperforming conventional design methods. The results indicate that the proposed approach not only enhances the resilience of RC structures under seismic events but also contributes to the development of intelligent structural control systems. This research underscores the potential of combining advanced control theories with artificial intelligence techniques to address contemporary challenges in structural engineering, paving the way for more resilient and adaptive building designs in earthquake-prone regions.
Key Words
Deep Reinforcement Learning (DRL); fuzzy model; hybrid heuristic search algorithm; seismic performance, sustainable and resilience; vibration mitigation
Address
(1) ZY Chen, Ruei-Yuan Wang, Yahui Meng:
School of Science, Guangdong University of Petrochemical Technology, Maoming, Guangdong, China;
(2) Huakun Wu:
School of Computer Science, Guangdong Polytechnic Normal University, Guangzhou, Guangdong, China;
(3) Timothy Chen:
Division of Engineering and Applied Science, Caltech, CA 91125, USA.
Abstract
Impedance-based damage detection technique has been pointed out as a useful method for structural health monitoring. However, it is needed to consider the temperature effect on the impedance signature to apply this technique in real structures. Also, for practical application, it is advantageous to consider the operating conditions such as vibration and live load. And, the correlation between the measured impedance data and the analysis results can improve the capability of structural health monitoring. In this study, an impedance-based damage detection method considering the temperature effect and operating conditions is experimentally researched for an aluminum beam, and the correlation between the impedance damage indices obtained from experiment and the equivalent bending stiffness for the cracked beam calculated by analysis is investigated. The long-term measurement under vibration is carried out, and it is found that the damage can be clearly estimated, while monitoring the warning criterion together, after compensating the temperature effect. It is also found that as the crack size increases, the variation in the damage indices increases, and the correlation between the impedance damage index due to crack propagation and the analyzed equivalent bending stiffness of the cracked beam is partially identified.
Key Words
damage detection; impedance; operating condition; practical application; temperature effect
Address
Department of Architectural Engineering, Namseoul University, 91 Daehak-ro, Seobuk-gu, Cheonan-si, Chungcheongnam-do 31020, Republic of Korea.
Abstract
For airplanes, especially aging airplanes, corrosion is widespread. As the corrosion reaches a certain level, it may cause flight accidents. Lamb wave tomography (LWT), as one of the Structural Health Monitoring technologies, can be used to monitor the corrosion of aircraft structures. However, the LWT requires densely arranged sensors on both sides of the monitoring area and has the poor imaging quality. These disadvantages limit its application in aircraft corrosion monitoring. In view of this situation, this paper proposes Bayesian Compressed Sensing (BCS)-based tomographic method to monitor corrosion of aircraft structure. BCS-based tomographic method reduces the number of sensors by under-sampling the received lamb wave signal and utilizes a Bayesian formulation to perform the original signal reconstruction. Compared to conventional LWT, the new method has better imaging quality with fewer sensors. Compared to the improved Compressed Sensing (CS)-based tomographic, BCS-based tomographic has fewer imaging artifacts, and shorter imaging time. Simulation and experiment are carried out on aviation aluminum plate with corrosion to verify the new proposed method. The results show the advantages of the proposed BCS-based tomographic method.
Address
(1) Zengnian Xin:
Research Center of Structural Health Monitoring and Prognosis, State Key Laboratory of Mechanics and Control for Aerospace Structures, Nanjing University of Aeronautics and Astronautics, 29 Yudao Street, Nanjing, Republic of China;
(2) Ming Chang:
School of Electrical, Energy and Power Engineering, Yangzhou University, 88 South University Road, Yangzhou, Republic of China.
Abstract
As the span increases, the difficulty of bridge construction control continuously escalates. Accurate construction control effectively ensures that bridges maintain a reasonable stress state, proper alignment, and track smoothness. This work innovatively integrates the Mean Value Theorem Expansion Response Surface method with a Neural Network Surrogate Model to precisely identify key parameters during the construction process, achieving high-accuracy predictions of construction alignment for large-span bridges. Initially, the Response Surface-Monte Carlo method is used for the sensitivity analysis of the main construction parameters. Subsequently, a parameter identification model is established to identify and correct key parameters affecting alignment and to refine the finite element model. Based on the adjusted model, sample data are collected to create an alignment prediction network model, which predicts alignment deviations for subsequent beam segments in construction, achieving high-precision reliability assessment of bridge construction alignment. The applications of case project demonstrate that the proposed methods for structural parameter identification and alignment prediction significantly enhance the precision of alignment forecasts. Characterized by the simplicity and high accuracy of the proposed method, it can offer a novel, efficient approach for alignment control under complex construction conditions.
Key Words
alignment prediction; construction control; large-span bridge; neural network surrogate model; response surface method; sensitivity analysis
Address
(1) Xingwang Sheng, Xu Song, Weiqi Zheng, Huanzhong Sun, Yonghong Yang:
School of Civil Engineering, Central South University, Changsha, Hunan 410075, China;
(2) Weiqi Zheng:
National Engineering Research Center for High-speed Railway Construction Technology, Changsha, Hunan 410075, China;
(3) Huanzhong Sun:
China Railway 14th Bureau Second Engineering Co., LTD., Tai'an, Shandong, China 271000, China
(4) Yonghong Yang:
Shanghai-Hangzhou Railway Passenger Dedicated Line Co., LTD., Shanghai, China 200040, China.
Abstract
Concrete core in concrete-filled steel tubes (CFSTs) is heterogeneous at mesoscale and the heterogeneity affects stress wave propagation within CFSTs. Coupling homogenization finite element models (CHFEMs) corresponding to traditional coupling mesoscale finite element models (CMFEMs) of rectangular CFSTs (RCFSTs) with randomly distributed elliptical and polygonal aggregates are established. The influences of concrete core heterogeneity and an interface debond defect on stress wave field and the responses of piezoelectric-lead-zirconate-titanate (PZT) sensors at different measurement distances are distinguished using homogenization finite element models (HFEMs) and the CHFEMs, respectively. A comparative test on thirty RCFST cross sections with and without an interface debond defect is performed to quantitatively evaluate the effect of concrete core heterogeneity on the responses of PZT sensors at different measurement distances and to distinguish it with that of the designed interface debond defect. Both mesoscale homogenization numerical simulation and test results show that the heterogeneity and mesoscale randomness of concrete core locally affect the response of PZT sensors that are close to the PZT actuator mounted on the surface of the steel tube of the CFSTs at certain levels. The influence of the interface debond defect on the stress wave fields within the cross sections of RCFSTs and the response of PZT sensors with measurement distances over 160 mm is dominant. The findings further illustrate the rationality of the interface debond defect detection method using stress wave measurement of PZT sensors with a suitable measurement distance for RCFSTs even concrete core in RCFSTs in practice is a heterogenous material with randomly distributed aggregates of different shapes at mesoscale.
Key Words
acoustic; concrete/reinforced concrete; dynamic analysis; experiment; finite element method; nondestructive evaluation; numerical material modelling; piezoelectric sensors and actuators; time domain
Address
(1) Jiang Wang, Bin Xu, Qian Liu, Ruiqi Guan:
College of Civil Engineering, Huaqiao University, Xiamen 361021, China;
(2) Bin Xu, Ruiqi Guan:
Key Laboratory for Intelligent Infrastructures and Monitoring of Fujian Province (Huaqiao University), Xiamen 361021, China;
(3) Xiaoguang Ma:
Foshan Graduate School, Northeastern University, Foshan 528311, China;
(4) Genda Chen:
Department of Civil, Architectural and Environmental Engineering, Missouri University of Science and Technology, Rolla, MO 65401, USA;
(5) Genda Chen:
INSPIRE University Transportation Center, College of Engineering and Computing, Missouri University of Science and Technology, Rolla, MO 65401, USA.
