Article
Engineering, Civil
Shixue Liang, Yuanxie Shen, Xiaodan Ren
Summary: This study proposes machine learning models to predict punching shear resistance and failure mode of reinforced concrete slab-column structures. Compared with other models, extreme gradient boosting (XGBoost) is selected as the best model, and its prediction process is explained using shapley additive explanation (SHAP) and partial dependence plot (PDP). The study reveals the relationships between punching shear resistance/failure and influential factors, which cannot be quantified through traditional experimental and theoretical analysis. Suggestions for choosing influential factors are provided to improve the punching shear resistance of RC slab-column joints.
Article
Multidisciplinary Sciences
Jianxin Zhang, Xiya Zhao, Yafei Gao, Wenye Guo, Yueyang Zhai
Summary: In this study, three efficient artificial neural network methods (back-propagation neural network, radial basis function neural network, and generalized regression neural network) were proposed to predict shear strengths and failure modes of beam-column joints. The results showed that these models are effective in predicting shear strength and failure mode. The generalized regression neural network model provided the most accurate results and is recommended for use in predicting shear strength and failure mode of beam-column joints in the structural design process.
ARABIAN JOURNAL FOR SCIENCE AND ENGINEERING
(2023)
Article
Engineering, Mechanical
Yuanxie Shen, Linfeng Wu, Shixue Liang
Summary: In this paper, an accurate prediction model is established to identify the failure mode of flat slabs. XGBoost is selected as the best model, and its prediction results are explained using SHAP.
ENGINEERING FAILURE ANALYSIS
(2022)
Article
Biology
Ke Wang, Jing Tian, Chu Zheng, Hong Yang, Jia Ren, Yanling Liu, Qinghua Han, Yanbo Zhang
Summary: This study established an ML model to accurately assess and stratify the risk of 3-year all-cause mortality in patients with heart failure caused by coronary heart disease. By combining ML with the SHAP method, individualized risk prediction was explained explicitly, providing physicians with an intuitive understanding of the influence of key features in the model.
COMPUTERS IN BIOLOGY AND MEDICINE
(2021)
Article
Engineering, Marine
Zihao Ding, Shixiong Zheng, Chuanhe Lei, Hongyu Jia, Zhiqiang Chen, Bo Yu
Summary: This study aimed to predict the residual horizontal bearing capacity and failure modes of corroded reinforced concrete (RC) columns in marine environments under vertical and seismic loads using machine learning (ML) methods. Six efficient ML algorithms were employed to establish predictive models, and the importance of input characteristic parameters was ranked using SHapley Additive exPlanations (SHAP) method. The study also investigated the influence of stirrup corrosion level, shear-span ratio, and axial load ratio on the residual bearing capacity and failure mode evolution of corroded RC columns. The proposed ML models showed great potential for accurately predicting the behavior of corroded RC columns.
Article
Engineering, Civil
Myeong-Ho Choi, Chang-Hwan Lee
Summary: A cyclic lateral load experiment was conducted to evaluate the performance of reinforced concrete beam-column joints with non-seismic detailing. Eight full-scale joint specimens were prepared with varying eccentricities, presence of slabs, and number of transverse beams. Most specimens exhibited joint shear failure after beam yielding, with reduced shear strength due to eccentricity. However, the presence of slabs and increased number of transverse beams improved the confinement and enhanced the joint shear strength. The calculated joint shear strength based on ASCE 41-17 showed significant discrepancy from actual measurements, highlighting the need for considering the impact of slabs and transverse beams to improve performance prediction. Future standards should also consider other factors that affect joint behavior based on further research outcomes.
Article
Biology
Amiel Meiseles, Denis Paley, Mira Ziv, Yarin Hadid, Lior Rokach, Tamar Tadmor
Summary: Chronic lymphocytic leukemia (CLL) is a common type of leukemia that primarily affects the elderly population. Predicting the necessity of treatment for CLL is important due to the heterogeneous behavior of the disease. This study aimed to develop a machine learning model to predict whether a patient will require treatment for CLL within two years of diagnosis based on demographic data and routine laboratory tests. The results showed that machine learning models outperformed the traditional prognostic scoring system in predicting the need for treatment.
COMPUTERS IN BIOLOGY AND MEDICINE
(2022)
Article
Construction & Building Technology
Benoit Hilloulin, Abdelhamid Hafidi, Sonia Boudache, Ahmed Loukili
Summary: This study introduces a machine learning framework for predicting the expansion of cementitious materials due to external sulfate attack (ESA). Four optimized machine learning models are compared and extreme Gradient Boosting (XGBoost) showed the best performance. SHapley Additive exPlanations (SHAP) enabled the identification of the most influential inputs and their relative influence. The model has been shown to accurately predict the time required to reach a given expansion.
JOURNAL OF BUILDING ENGINEERING
(2023)
Article
Environmental Sciences
Abolfazl Abdollahi, Biswajeet Pradhan
Summary: One of the worst environmental catastrophes in Australia is wildfire. Machine learning algorithms are used to identify fire occurrence patterns and susceptibility in wildfire-prone regions. The Shapley additive explanations model is used to interpret the results of a deep learning model for wildfire susceptibility prediction, revealing the significant contributions of factors such as humidity, wind speed, rainfall, elevation, slope, and NDMI.
SCIENCE OF THE TOTAL ENVIRONMENT
(2023)
Article
Engineering, Civil
Hosein Naderpour, Masoomeh Mirrashid, Payam Parsa
Summary: This article introduces a new efficient method for classifying failure modes in reinforced concrete columns using machine learning techniques. Through a comparison study, the decision tree model is found to provide desirable accuracy and computational simplicity in determining the failure mode.
ENGINEERING STRUCTURES
(2021)
Article
Engineering, Civil
Tadesse G. Wakjira, Mohamed Ibrahim, Usama Ebead, M. Shahira Alam
Summary: This paper presents a data-driven approach to determine the load and flexural capacities of RC beams strengthened with FRCM composites. Several machine learning models are evaluated to propose the best predictive model. The xgBoost model is found to be the most accurate and robust. A comparative study with existing analytical models shows the superiority of the proposed model. Additionally, an explanation approach and reliability analysis are used to interpret and validate the model.
ENGINEERING STRUCTURES
(2022)
Article
Computer Science, Interdisciplinary Applications
Xu Zhang, Lin Liu, Minxuan Lan, Guangwen Song, Luzi Xiao, Jianguo Chen
Summary: The interpretability of advanced machine learning models is utilized in this study to overcome the limitation of estimating variable contributions in crime prediction. Based on routine activity theory and crime pattern theory, 17 variables are selected for crime prediction, and the Shapley additive explanation (SHAP) method is used to discern the contribution of individual variables. Findings reveal that the proportion of the non-local population and the ambient population aged 25-44 contribute the most to crime prediction. Local models provide insights for tackling important factors at each location, while the global model identifies essential factors for the entire region.
COMPUTERS ENVIRONMENT AND URBAN SYSTEMS
(2022)
Article
Materials Science, Ceramics
XiaoWei Liu, ZhiLin Long, Li Peng
Summary: Amorphous alloys are formed by rapid cooling of liquid alloys, showing excellent mechanical properties due to the absence of grain boundaries, dislocations, and other defects, with Vickers hardness (HV) being one of its outstanding properties. In this study, four machine learning models were applied for HV modeling using atomic fraction, structural features, and load as input features. The Light Gradient Boosting Machine (LightGBM) model achieved higher determination coefficients R2 of 0.981 and 0.979 in the test set compared to the other three models, indicating superior generalization ability. Additionally, the introduction of shapley additive explanations (SHAP) theory improved the interpretability of the model, highlighting XP1 and Tm1 as the two most important features with critical values for improving HV if they fall within the correct range.
JOURNAL OF NON-CRYSTALLINE SOLIDS
(2023)
Article
Construction & Building Technology
Syed Farasat Ali Shah, Bing Chen, Muhammad Zahid, Muhammad Riaz Ahmad
Summary: Alkali activated material (AAM) or geopolymer has become a sustainable alternative to cement due to its low power consumption and greenhouse gas emissions, as well as good mechanical and durability features. However, developing AAM mixtures with desired properties is challenging due to the nature and diversity of available source materials. This study evaluates the performance of various machine learning models for predicting the compressive strength of one-part AAM binder, with XGBoost outperforming other algorithms. The use of SHapley Additive exPlanations (SHAP) helps interpret the predicted compressive strength and evaluate the effects of various parameters. The interpretable ML strategy used in this study aids in the production and performance tuning of durable and sustainable one-part AAMs for widespread applications.
CONSTRUCTION AND BUILDING MATERIALS
(2022)
Article
Construction & Building Technology
Gideon A. Lyngdoh, Mohd Zaki, N. M. Anoop Krishnan, Sumanta Das
Summary: This paper discusses the prediction of the non-linear composition-strength relationship in concretes using machine learning. It addresses the challenges of incomplete datasets and the complexity in interpreting results. The developed approach combines data imputation, machine learning, and data interpretation to evaluate the composition-strength relationship in concrete efficiently.
CEMENT & CONCRETE COMPOSITES
(2022)
Article
Engineering, Mechanical
Willy Ank de Morais, Railson Bolsoni Falcao, Mario Boccalini Jr, Fernando Jose Gomes Landgraf
Summary: This study compares the fatigue behavior of bcc Nb-48 wt%Ti (Ti-36at%Nb) alloy parts processed by Laser Powder Bed Fusion (L-PBF) using two different types of powders (spherical powder produced by Plasma Atomization and irregularly shaped powder produced by Hydride-Dehydride). The results show that the fatigue behavior is similar for both powders, but the specimens produced from the Hydride-Dehydride powder have higher dispersed results. Fracture primarily occurs in the porosities near the machined sample surface, and the fatigue cracks propagate in a zigzag mode parallel to the L-PBF building direction.
ENGINEERING FAILURE ANALYSIS
(2024)
Article
Engineering, Mechanical
Guoxi Jing, Shuai Sun, Teng Ma, Shubo Li, Tian Ma, Junchao Wei, Jianchao Pang
Summary: This study proposes a parameter identification method for predicting the TMF life of CGI materials used in cylinder heads. The study evaluates the TMF behavior of RuT450 material and successfully predicts its TMF life using multi-objective optimization. The results highlight the influence of plastic strain amplitude, temperature, and strain rate on the damage mechanism and life prediction.
ENGINEERING FAILURE ANALYSIS
(2024)
Article
Engineering, Mechanical
Xin Zhang, Xu Li, Weipeng Xu, Kuidong Gao, Kao Jiang, Xinyu Wang, Hongxin Wei
Summary: Conical picks are commonly used in tunnel excavation and often experience high temperatures during the cutting process. This study investigated the influence of temperature on the wear of conical picks and proposed a method for measuring rock abrasiveness that takes into account the thermal effect. The experimental results showed that cooling liquid reduced the maximum temperature and the mean mass loss of the conical pick. Additionally, the study examined the characteristics of oxygen and found evidence of peeling, cracking, and oxidation on the grinding rod used for rock abrasiveness measurement.
ENGINEERING FAILURE ANALYSIS
(2024)
Article
Engineering, Mechanical
Vivek Srivastava, Deependra Singh, A. G. Rao, V. P. Deshmukh
Summary: Engine gear failures are a significant concern in various industries. Investigating each gear failure and determining the root cause is important for preventing future failures. This study used experimental and numerical analysis techniques to investigate the premature failure of a flywheel gear in a marine diesel engine. The findings revealed that the gear failed due to overload stresses caused by overload torque generated from sudden inertial thrust in the internal combustionengine.
ENGINEERING FAILURE ANALYSIS
(2024)
Article
Engineering, Mechanical
Jun Zhang, Weidong Liu, Feilong Liu, Zhiqiang Yang, Jinfang Peng, Minhao Zhu
Summary: During a major overhaul of an aeroengine, a cracked stator vane of the low-pressure compressor was discovered, with the crack originating from a corrosion pit located on the arc transition between the leading edge and the upper edge plate. The fracture surface exhibited signs of fatigue cracks, indicating a pit-related fatigue crack. Metallographic examination revealed that most corrosion pits were distributed beneath the bulging or peeling paint layer on the leading edge and the concave side. Identification of the corrosion products showed the presence of sulfate and alumina, suggesting atmospheric corrosion as a likely cause for the cracked stator vane.
ENGINEERING FAILURE ANALYSIS
(2024)
Article
Engineering, Mechanical
Juan Jose Jimenez de Cisneros Fonfria, Ester Olmeda, Susana Sanz, Maria Garrosa, Vicente Diaz
Summary: Energy absorbing devices in vehicles can partially transform impact energy and protect passengers. In this paper, the failure analysis of the underframe crumple zone in a train car was conducted through finite element simulations, showing that the component acted properly after the impact.
ENGINEERING FAILURE ANALYSIS
(2024)
Article
Engineering, Mechanical
Fengqi Zhang, Gang Luo, Haiyang Zhang, Peihong Cong, Lulu Liu, Wei Chen
Summary: In this study, a predictive model for bird strike trajectory was developed based on the quasi-linear viscoelastic model. By combining simulation analysis under multiple conditions, the model reproduced the trajectory change law in low and medium speed bird impact experiments.
ENGINEERING FAILURE ANALYSIS
(2024)
Article
Engineering, Mechanical
S. M. de Gouveia, L. de Abreu Correa, D. B. Teles, M. Oliveira, T. G. R. Clarke
Summary: Emergency Shutdown Valves (EDSVs) are used in industrial applications to interrupt fluid flow during hazardous events. The damage to the valve seats can be detected by analyzing the pressure and torque data. This study compared three processing options and found that evaluating the complete signature with a Gradient Boosting Classifier algorithm is the most effective strategy.
ENGINEERING FAILURE ANALYSIS
(2024)
Article
Engineering, Mechanical
Xinyan Jin, Kai Wang, Hongwei Qian
Summary: This study investigated the root cause of an abnormal premature groove-clogging failure on a sink roll in a hot-dip galvanizing line. It was found that the failure was caused by the buildup of a large amount of Fe2Al5Znx dross in the grooves, and the key factor was the control of Tstrip.
ENGINEERING FAILURE ANALYSIS
(2024)
Article
Engineering, Mechanical
Zhiqiang Li, Jie Li, Cen Li, Xiaodong Xie, Zhiyong Yang
Summary: In this study, the microstructure evolution of the material in the hot spot area of the high-speed train brake disc was observed and the mechanism of mechanical property degradation was obtained through tensile tests. The research results provide guidance for crack damage identification of the brake disc and condition-based maintenance.
ENGINEERING FAILURE ANALYSIS
(2024)
Article
Engineering, Mechanical
Lijia Luo, Yincheng Wang, Wei Chen, Zuming Zhao, Wenfei Chen, Shiyi Bao
Summary: Tube-to-tube impact wear is a significant factor in the failure of alloy 690 tubes in nuclear steam generators. This paper investigates the wear damage mechanism and detection method of alloy 690 tubes under tube-to-tube impact loads. The results reveal the wear damage mechanism through analysis of macro and micro morphologies as well as metallographic structures. A nonlinear ultrasonic wave mixing method is proposed for damage detection and a detection device is developed to implement this method. Experimental results indicate that the damage mode of tube-to-tube impact wear varies with impact load and cycle number, and the nonlinear modulation index is effective in assessing the damage degree of alloy 690 tubes.
ENGINEERING FAILURE ANALYSIS
(2024)
Article
Engineering, Mechanical
Xing-Ju Yang, Feng Lin, Xu Du, Lu Qiu
Summary: This paper proposes a novel method of improving the progressive collapse resistance of RC frame structures by adding locally debonded kinked steel plates (KPs). Experimental tests and finite element analysis reveal that the addition of KPs significantly enhances the ultimate resistance and deformability of the beam-slab substructures. The mechanism of resistance improvement is explained, and an analytical model is developed for predicting the ultimate resistance of the structures.
ENGINEERING FAILURE ANALYSIS
(2024)
Article
Engineering, Mechanical
Xiaohe Wang, Zengqiang Cao
Summary: This paper presents a dynamic installation (DI) method based on electromagnetic force to reduce damage to composite laminates and improve joint performance. The study shows that DI can effectively decrease installation resistance, reduce severe damage to laminates, and improve stress distribution in the joint.
ENGINEERING FAILURE ANALYSIS
(2024)
Article
Engineering, Mechanical
Jitong Cui, Yingying Kong, Cuiwei Liu, Baoping Cai, Faisal Khan, Yuxing Li
Summary: This paper presents a method for quantifying the failure probability of hydrogen doped pipelines and utilizes Bayesian network for quantitative calculation. The results show that this method can more accurately assess the failure probability of hydrogen doped pipelines and provide a decision basis for preventing failure accidents.
ENGINEERING FAILURE ANALYSIS
(2024)
Article
Engineering, Mechanical
Yiping Jiang, Maoru Chi, Jungang Yang, Liangcheng Dai, Yuchen Xie, Zhaotuan Guo
Summary: This article investigates and simulates the derailment of empty freight trains at the switch section of a turnout in the diverging route, and proposes preventive measures. The results show that the derailment is caused by the combined effect of various factors, and reducing the longitudinal coupler force and the wheel rail friction coefficient can decrease the possibility of derailment. Additionally, installing a guard rail in front of the turnout can effectively prevent derailment.
ENGINEERING FAILURE ANALYSIS
(2024)
