Techno Press
Tp_Editing System.E (TES.E)
Login Search
You logged in as...

sss
 
CONTENTS
Volume 36, Number 1, July 2025
 

Abstract
In the field of structural engineering, traditional design and optimization methodologies are being transformed by the integration of machine learning (ML) and deep learning (DL) techniques. The increasing complexity of structural systems driven by variations in geometric parameters such as eave height, span length, frame spacing, and support conditions, as well as the growing demand for cost-effective solutions and reduced design computation time, has prompted engineers to adopt innovative approaches, including predictive algorithms for steel section selection. A comprehensive dataset was generated using finite element modeling (FEM) to represent a wide range of structural configurations. These configurations were validated through manual calculations to ensure data accuracy for ML and DL training. The trained models analyze structural parameters to predict optimal section dimensions, addressing nuanced design requirements. Various ML algorithms, including Polynomial Regression, LightGBM, XGBoost, Random Forest, and artificial neural networks (ANN) were employed for predicting column and beam dimensions. Among them, Random Forest achieved the highest accuracy (Adjusted R2 = 94.132%, MAE = 0.336), followed by ANN (Adjusted R2 = 91.63%, MAE = 0.212). A graphical user interface (GUI) was also developed to bridge the gap between model predictions and practical implementation, enabling engineers to design cost-effective and resilient structures in compliance with AISC, ASCE, and ECP standards.

Key Words
algorithms; ANN; data base; deep learning; frames; machine learning; optimization; steel

Address
Cairo University, Faculty of Engineering, Structural Engineering Department, Giza, 12613, Egypt.


Abstract
This study aims to develop a 2D CNN deep learning model processing electromechanical impedance (EMI) responses of a capsule-like smart aggregate (CSA) sensor for monitoring stress variation in concrete structures. The following approaches are conducted to obtain the objective. Firstly, an overall scheme of the proposed method is presented. An EMI measurement model is theoretically presented for a CSA sensor embedded in a concrete cylinder under compressive loadings. A 2D CNN model is designed to learn and classify stress-sensitive features from CSA's EMI responses. Secondly, a CSAembedded concrete cylinder is experimentally investigated to record the EMI signals of the cylinder under a series of compressive stress levels. Thirdly, the performance of the 2D CNN model is investigated for noise-contaminated data sets as well as untrained stress-EMI scenarios. Finally, the accuracy of the proposed 2D CNN model is analyzed by comparatively discussing with a well-established 1D CNN model.

Key Words
compressive testing; concrete structure; convolutional neural network; impedance-based method; PZT sensor; smart aggregate; stress and damage monitoring

Address
(1) Quoc-Bao Ta, Ngoc-Lan Pham, Jeong-Tae Kim:
Department of Ocean Engineering, Pukyong National University, 45 Yongso-ro, Nam-gu, Busan 48513, Republic of Korea;
(2) Quang-Quang Pham:
Bridge and Road Department, Danang Architecture University, Da Nang 550000, Viet Nam.

Abstract
Delay compensation presents a key component to ensure an accurate and robust real-time hybrid test (RTHT). As RTHT continues to evolve, the scale and degrees of freedom of studied structures increase, requiring greater attention to the responses of higher-order modes. High-frequency components in structural responses at the boundaries of substructures must also be accurately imposed onto the specimen. Robust delay compensation for high-frequency signals loading in RTHT needs to be explored. This paper proposes an adaptive time delay compensation based discrete model strategy combined with sliding mode control method (ADM-SMC). In this method, combining with the robust sliding mode control (SMC), the least squares method is used to update the parameters of the discrete model of the SMC-loading system in real time, which provides a robust compensation method for the wide-band and high-frequency excitation loading. A delay compensation test and a RTHT were conducted to verify the feasibility of this method. The results indicate that the proposed method can compensate varying time delay precisely with robustness for different high-frequency excitation loading.

Key Words
adaptive delay compensation; high-frequency signal; real-time hybrid testing; sliding mode control

Address
(1) Yunhai Zeng, Yucai Chen, Ping Tan:
School of Civil Engineering and Transportation, Guangzhou University, Guangzhou 510006, China;
(2) Huimeng Zhou:
Earthquake Engineering Research & Test Center, Guangzhou University, Guangzhou 510006, China;
(3) Yu Guo, Zhen Wang:
School of Civil Engineering and Architecture, Wuhan University of Technology, Wuhan 430070, China.

Abstract
Efficient and accurate ground motion simulation is key to achieving reliable structural seismic analysis. However, ground motions are significantly influenced by multiple factors, such as the source mechanism, propagation path, and site conditions, resulting in notable nonstationarity and spatial variability. Traditional ground motion synthesis methods based on statistical models are often inadequate to address these complexities. Therefore, this paper proposes a ground motion synthesis method based on Kernel Principal Component Analysis (KPCA) and Genetic Algorithm (GA). By extracting characteristic mother waves from specific earthquake records and using genetic algorithms, artificial ground motions that meet the target response spectrum and peak ground acceleration (PGA) are generated. The method is applied to the Hualien earthquake in Taiwan as an example, and the results show that the synthesized ground motions effectively match the target response spectrum and PGA, with a maximum error within 7%. Further power spectral analysis demonstrates a high similarity between the synthesized and natural ground motions in terms of power spectral characteristics. Moreover, a finite element analysis of a typical frame structure under both artificial and similar natural ground motions from a database shows consistent seismic responses and patterns between the two. To further validate the applicability of the proposed method, ten ground motions were removed from the database, and artificial ground motions were synthesized based on their acceleration response spectra and peak ground velocity (PGV) information. Comparisons of waveform characteristics, PGA, PGV, and acceleration response spectra indicate that the artificial ground motions effectively retain the waveform characteristics of natural ground motions. The proposed method provides an efficient tool for engineering seismic applications, capable of synthesizing artificial ground motions that preserve the waveform and damage characteristics of natural ground motions while meeting specified target response spectrum, PGV, PGA, and other parameters.

Key Words
genetic algorithm; ground motion synthesis; Kernel Principal Component Analysis (KPCA); nonstationarity; response spectrum; seismic simulation

Address
(1) Zhen Liu:
School of Management Science and Engineering, Shandong Technology and Business University, No. 191, Binhai Middle Road, Laishan District, Yantai City, Shandong, China;
(2) Xingliang Ma:
School of Civil Engineering and Architecture, Changzhou Institute of Technology,

Techno-Press: Publishers of international journals and conference proceedings.       Copyright © 2026 Techno-Press ALL RIGHTS RESERVED.
P.O. Box 33, Yuseong, Daejeon 34186 Korea.
General Inquiries: info@techno-press.com / Journal Administration: admin@techno-press.com