Engineering Applications of Artificial Intelligence最新文献

{"title":"Numerical study of tri-hybrid nanofluids in a rectangular cavity with an enclosed circle via COMSOL and Levenberg-Marquardt method","authors":"Sami Ul Haq ,&nbsp;Arooj Tanveer ,&nbsp;Muhammad Bilal Ashraf ,&nbsp;Nidhal Becheikh ,&nbsp;Kaouther Ghachem ,&nbsp;Lioua Kolsi","doi":"10.1016/j.engappai.2026.114089","DOIUrl":"10.1016/j.engappai.2026.114089","url":null,"abstract":"<div><div>This work focuses on a numerical simulation of magneto mixed convection transport in electrically conducting tri hybrid nanofluids that is enclosed in a two dimensional rectangular lid driven cavity with a cold circular obstacle. Dissipative processes due to viscous dissipation and Joule heating are taken into account and the non-dimensional governing equations are resolved by Galerkin finite-element method in COMSOL Multiphysics. The effects of the major controlling parameters, i.e. the Hartmann number <span><math><mrow><mo>(</mo><mrow><mn>0.1</mn><mo>≤</mo><mi>M</mi><mo>≤</mo><mn>20</mn></mrow><mo>)</mo></mrow></math></span>, Reynolds number <span><math><mrow><mo>(</mo><mrow><mn>100</mn><mo>≤</mo><mi>R</mi><mi>e</mi><mo>≤</mo><mn>500</mn></mrow><mo>)</mo></mrow></math></span>, the Richardson number <span><math><mrow><mo>(</mo><mrow><mn>0.1</mn><mo>≤</mo><mi>R</mi><mi>i</mi><mo>≤</mo><mn>10</mn></mrow><mo>)</mo></mrow></math></span>, and the nanoparticle volume-fraction coefficients (<span><math><mrow><mn>0</mn><mo>≤</mo><mo>∅</mo><mo>≤</mo><mn>0.06</mn></mrow></math></span>), on the flow structure and heat-transfer characteristics are systematically evaluated. These findings indicate that <em>Ha</em> increase inhibits fluid motion by the force of Lorentz forces, thus minimising convective exchange of heat at the moving heated wall. On the other hand, increased values of Re significantly increase fluid flow and thermal mixing resulting in increased local and mean Nusselt numbers. Tri-hybrid nanoparticles <span><math><mrow><mo>(</mo><mrow><mi>A</mi><mi>u</mi><mo>,</mo><mi>A</mi><mi>g</mi><mo>,</mo><mi>T</mi><mi>i</mi><msub><mi>O</mi><mn>2</mn></msub></mrow><mo>)</mo></mrow></math></span> enhance the thermal capability of the base fluid by increasing the effective thermal conductivity, thereby, enhancing the overall heat-transfer rate in the base fluid. A high Richardson number works the flow field in the direction of buoyancy-dominated convection, dampens the contribution of forced-convection, and reduces the transfer of heat to that moving away of the upper moving wall. It uses an artificial neural network that has been trained using the Levenberg-Marquardt algorithm to forecasts and optimise the average Nusselt number, with excellent correspondence with computed data; the regression coefficient approaches one, and the mean squared error is small. High Reynolds number, low Hartmann number, low Richardson number, and moderate volume fractions of nanoparticle yield the best results with regard to heat-transfer performance, and thus, the study irrevocably supports the capability of integrating tri-hybrid nanofluids with data-driven optimization in the context of advanced thermal-management operations.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"170 ","pages":"Article 114089"},"PeriodicalIF":8.0,"publicationDate":"2026-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146161877","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Machine learning-based prediction of ductility of strain-hardening fiber-reinforced cementitious composites","authors":"Tan Duy Phan,&nbsp;Van Thong Nguyen,&nbsp;Dong Joo Kim","doi":"10.1016/j.engappai.2026.113915","DOIUrl":"10.1016/j.engappai.2026.113915","url":null,"abstract":"<div><div>The high ductility, characterized by both strain capacity and average crack spacing, of strain-hardening fiber-reinforced cement composites (SH-FRCCs) is expected to enhance the load-carrying capacity and durability of buildings and infrastructure made of SH-FRCCs. This study aimed to predict the strain capacity and average crack spacing of SH-FRCCs using four popular machine learning (ML) models: k-nearest neighbor (k-NN), decision tree (DT), random forest (RF), and adaptive boosting (ADB) models. Nine input variables, the matrix compressive strength, fiber type 1, tensile strength of fiber 1, fiber type 2, tensile strength of fiber 2, fiber index, specimen width, specimen thickness, and gauge length were considered in the ML models. Among the investigated ML models, the RF model exhibited relatively good performance in predicting the strain capacity (R² = 0.986) and log-transform crack spacing (R² = 0.955) of SH-FRCCs in training data. The better performance of the RF model is attributed to the model's ensemble structure, which integrates multiple decision trees, effectively reduces variance, and manages complex data structures. Fiber index is the most influential variable on both strain capacity and average crack spacing of SH-FRCCs, based on SHapley Additive Explanations (SHAP) and Partial Dependence Plots (PDP) analysis. The strain capacity of SH-FRCCs decreased with increasing specimen width and gauge length, whereas crack spacing increased with specimen width. Finally, the developed ML model was validated against experimental data, showing excellent agreement with deviations below 10 % for strain capacity and around 11 % for average crack spacing.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"170 ","pages":"Article 113915"},"PeriodicalIF":8.0,"publicationDate":"2026-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146193012","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A communication-efficient federated learning method for traffic flow prediction","authors":"Kaiju Li ,&nbsp;Qiang Xu ,&nbsp;Dong Wang ,&nbsp;Xiang Nie ,&nbsp;Hao Wang","doi":"10.1016/j.engappai.2026.114182","DOIUrl":"10.1016/j.engappai.2026.114182","url":null,"abstract":"<div><div>Federated learning is increasingly adopted for traffic flow prediction (TFP) to enable privacy preserving collaboration across distributed sensors. However, real-world deployments are highly heterogeneous in computational capability, causing stragglers that dominate per-round latency and severely slow down model updates. Most existing approaches mitigate stragglers by suppressing or discarding slow clients, which reduce data representativeness and introduce training bias. It is a harmful trade-off for TFP where broad spatial coverage is crucial for accuracy. We propose a communication-efficient logical clustering federated learning framework (LCFed) that mitigates stragglers by logically balancing effective training time while preserving full client participation. LCFed combines a coarse-grained logical dynamic clustering algorithm (<span>LoDynClust</span>) to balance computational resources across clusters and reduce synchronization delays, with a fine-grained intra-cluster adaptive collaborative training mechanism (<span>ICACT</span>) to regulate aggregation intervals and mitigate training bias. We further provide a convergence analysis. Extensive experiments on three real-world traffic datasets show that LCFed significantly reduces training latency caused by stragglers while maintaining competitive prediction accuracy compared with state-of-the-art baselines.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"170 ","pages":"Article 114182"},"PeriodicalIF":8.0,"publicationDate":"2026-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146193286","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Separable physical spatiotemporal graph message aggregation for fault diagnosis","authors":"Kuangchi Sun ,&nbsp;Aijun Yin ,&nbsp;Yihua Hu","doi":"10.1016/j.engappai.2026.114109","DOIUrl":"10.1016/j.engappai.2026.114109","url":null,"abstract":"<div><div>Spatiotemporal graph has become a research hotspot for it can excavate spatiotemporal information in multi-sensor fault diagnosis. However, the existing methods do not fully consider the physical attenuation characteristics in edge when the fault features are transmitted to the next sensor in the case of cross-sensor spatial temporal correlation. Besides, existing spatiotemporal convolutional networks pay much attention to the integration of all nodes for information update and the network structure design without realize the aggregation of edge information with different attributes. To address these issues, we propose Separable Physical Spatiotemporal Graph Message Aggregation (SPSGMA) for Fault Diagnosis. Firstly, a spatiotemporal graph of physical connection properties across sensors is proposed to assign different properties to different edges. Then, a novel wavelet frequency selection method is proposed for node feature extraction of different physical edge. Finally, a separable message aggregation network is designed to realize aggregation of frequency messages on different physical edges and classification rather than unified feature extraction. Three different datasets are used to verify the effectiveness of SPSGMA. Compared with other methods, SPSGMA achieves the best diagnostic performance in long chain sensor data diagnosis, and its average diagnosis accuracy in different diagnosis respectively are 99.99%, 98.59%, and 99.93%.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"170 ","pages":"Article 114109"},"PeriodicalIF":8.0,"publicationDate":"2026-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146161801","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Multi-label financial statement fraud detection based on long short-term memory and multilayer perceptron hybrid model","authors":"Zhensong Chen ,&nbsp;Hao Chen ,&nbsp;Yanxin Liu ,&nbsp;Yong Shi","doi":"10.1016/j.engappai.2026.114188","DOIUrl":"10.1016/j.engappai.2026.114188","url":null,"abstract":"<div><div>The detection of financial statement fraud holds paramount importance due to its impact on economic order, public trust, and legal accountability. However, existing studies often treat it as a binary classification problem, while overlooking the valuable information from correlations between various fraud types. To address this issue, we propose a multi-label financial statement detection framework to identify all distinct types of fraudulent behaviors, including inflated profits, inflated assets, false statements, delay in disclosure, and omission of significant information. In the proposed framework, a Long Short-Term Memory (LSTM) network with a MultiLayer Perceptron (MLP) are integrated to effectively capture temporal dependencies and correlations among different types of fraud. Furthermore, we employ interpretability techniques to analyze the differences and connections between various types of fraud and different financial features. Empirical results on real Chinese datasets have demonstrated the effectiveness of the proposed multi-label classification framework, verified its superiority over traditional binary classification models, and maintained robustness in the case of class imbalance. In addition, we propose a novel Top-K thresholding strategy. Its core idea is to determine specific fraud types involved based on an initial assessment of the severity of its fraudulent behavior. Overall, this research contributes to the field of financial statement fraud detection by introducing a multi-label classification framework and conducting thorough interpretability analysis. This advancement not only provides auditors and regulators with actionable tools for more targeted investigations but also fosters more comprehensive understanding of the mechanisms underlying financial fraud.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"170 ","pages":"Article 114188"},"PeriodicalIF":8.0,"publicationDate":"2026-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146193016","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Neighborhood constrained attention for lightweight image super-resolution","authors":"Rui He ,&nbsp;Zhenyang Zhu ,&nbsp;Xiaoyang Mao","doi":"10.1016/j.engappai.2026.114119","DOIUrl":"10.1016/j.engappai.2026.114119","url":null,"abstract":"<div><div>In recent years, to improve image super-resolution performance, several studies have explored integrating convolutional modules with vision transformers (ViTs) to enhance the local feature modeling of ViTs. However, these hybrid approaches often introduce inconsistencies in feature representation, redundant information, and an increased number of parameters, ultimately limiting both performance and computational efficiency. To overcome these challenges, we propose a novel neighborhood constrained attention (NCA) mechanism that enables transformers to effectively capture both local and global features without requiring additional convolutional modules. Specifically, we first divide the window into a set of <span><math><mrow><mi>r</mi><mo>×</mo><mi>r</mi></mrow></math></span> grids, treating them as local features, and then explore both intra- and inter-relationships within and across these local features, using them as constraints to refine window attention. Furthermore, instead of relying on averaging or other heuristic schemes for assigning labels to local features, we combine them through a linear transformation, ensuring label accuracy and uniqueness. Extensive experiments demonstrate that the proposed NCA not only outperforms other state-of-the-art lightweight approaches on public benchmark datasets but also excels in engineering image datasets, such as automated defect detection and product quality inspection, while requiring fewer parameters and lower computational costs. Notably, compared to <span><math><mo>×</mo></math></span>4 SwinIR-light (SwinIR: Image Restoration Using Swin Transformer), NCA achieves an average performance gain of 0.28 dB across five public test sets while reducing network parameters by 27% and computational complexity (floating point operations, FLOPs) by 30%. Code and models are obtainable at <span><span>https://github.com/hms-source/NCA</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"170 ","pages":"Article 114119"},"PeriodicalIF":8.0,"publicationDate":"2026-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146193013","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Multiphysics response and internal leakage prediction of seismic hydraulic systems considering structural clearance effects","authors":"Donglai Li ,&nbsp;Jianying Li ,&nbsp;Tiefeng Li ,&nbsp;Xiaoyan Du","doi":"10.1016/j.engappai.2026.114140","DOIUrl":"10.1016/j.engappai.2026.114140","url":null,"abstract":"<div><div>Accurate prediction of internal leakage in valve controlled hydraulic systems remains challenging because of the strong nonlinear coupling among pressure, clearance, and transient flow behavior. This study develops a multiphysics modeling framework that integrates computational fluid dynamics (CFD) with a Physics constrained Kernel Additive Network for Leakage Prediction (PKAN-LP). A dynamic mesh simulation of the spool valve and hydraulic cylinder is employed to capture the influence of structural clearances on leakage evolution. By embedding residuals from the Navier-Stokes and Reynolds equations, the PKAN-LP framework enables structure driven learning and enhances prediction stability and physical consistency across both steady and transient regimes. The results demonstrate that PKAN-LP achieves accurate and physically coherent predictions, effectively mitigating leakage overshoot under high pressure and large clearance conditions. Shapley Additive Explanations (SHAP) based sensitivity analysis reveals that radial clearance is the dominant factor, followed by valve opening and inlet pressure, and their interactions govern the nonlinear leakage behavior. This study advances physics informed modeling of hydraulic systems by bridging data driven learning with physical interpretability, providing a generalizable framework for modeling and optimization of high-pressure hydraulic systems.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"170 ","pages":"Article 114140"},"PeriodicalIF":8.0,"publicationDate":"2026-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146193011","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Explainable artificial intelligence-Infused hybrid transfer learning framework with multiscale feature fusion for brain tumor detection and classification","authors":"Shahid Mohammad Ganie ,&nbsp;Rama Chaithanya Tanguturi ,&nbsp;Manahil Mohammed Alfuraydan","doi":"10.1016/j.engappai.2026.114128","DOIUrl":"10.1016/j.engappai.2026.114128","url":null,"abstract":"<div><div>Brain tumors represent a significant health issue and are a leading cause of cancer-related fatalities globally. Early detection and accurate classification approaches are essential for addressing this critical health issue. This study proposes a novel hybrid deep multiscale integration network (DMI-Net) model for brain tumor diagnosis using magnetic resonance imaging (MRI) dataset. Image preprocessing included resizing, contrast enhancement using contrast-limited adaptive histogram equalization (CLAHE), normalization, and Gaussian filtering to enhance image quality. A lightweight parallel depthwise separable convolutional neural network (PD-CNN) is designed to extract multiscale relevant features with minimum computational resources. Principal component analysis (PCA), linear discriminant analysis (LDA), uniform manifold approximation and projection (UMAP), and t-distributed stochastic neighbor embedding (t-SNE) were used to visualize and validate the class-separable structure of the feature space in interpretability assessment. The hybrid framework was developed by stacking and concatenating three top-performing transfer learning (TL) models and integrating them with the PD-CNN architecture. Evaluation was conducted using standard performance metrics. For interpretability in clinical decision-support, model outputs were analyzed using shapley additive explanations (SHAP) and gradient-weighted class activation mapping (Grad-CAM) and its variants. The DMI-Net model demonstrated superior results compared with eight TL models, achieving an accuracy of 99.24%, precision of 99.00%, recall of 98.42%, F1-score of 98.54%, and area under the receiver operating characteristic curve of 98.85%. It outperformed existing state-of-the-art studies in the literature. The results indicate the potential utility of the proposed model for increasing confidence in diagnosing brain tumors, supporting clinical decision-making.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"170 ","pages":"Article 114128"},"PeriodicalIF":8.0,"publicationDate":"2026-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146193288","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Prediction of flutter derivatives for closed-box bridge girder: A feature-fusion residual neural network algorithm","authors":"Chuanting Liu ,&nbsp;Genshen Fang ,&nbsp;Zuopeng Wen ,&nbsp;Ke Li ,&nbsp;Yaojun Ge","doi":"10.1016/j.engappai.2026.114142","DOIUrl":"10.1016/j.engappai.2026.114142","url":null,"abstract":"<div><div>Flutter derivatives are crucial parameters for aerodynamic performance analysis of long-span bridges, which are typically identified through time-consuming and costly methods such as wind tunnel tests or computational fluid dynamics (CFD). This study proposes a deep learning approach for the rapid identification of the flutter derivatives of closed-box girders, utilizing feature-fusion residual network architecture (FF-ResNet). We construct a dataset comprising flutter derivatives of 113 cross-sections at eight reduced wind speeds, and the flutter derivatives are identified via multi-frequency forced vibration CFD simulations. Then, the reduced wind speed and a pre-processed image of the cross-section are used as inputs, and the model is trained to learn multi-modal features. Bayesian optimization is employed to enhance predictive accuracy for flutter derivatives, with the model achieving r-squared (<span><math><mrow><msup><mi>R</mi><mn>2</mn></msup></mrow></math></span>) values exceeding 0.97 on the training set and 0.92 on the validation set; in 10-fold cross-validation, the average <span><math><mrow><msup><mi>R</mi><mn>2</mn></msup></mrow></math></span> of the validation set across ten folds also exceeds 0.92, demonstrating high accuracy. Next, the model is used to analyze the variation of flutter derivatives across the aerodynamic shape range, and the SHapley Additive exPlanations (SHAP) algorithm is applied to investigate the importance of the geometric parameters. The predicted flutter derivatives are then employed to compute the critical wind speed distribution over the range of considered cross-section variations.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"170 ","pages":"Article 114142"},"PeriodicalIF":8.0,"publicationDate":"2026-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146193285","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Spatio-temporal grey Bernoulli model for green development of marine economy forecasting","authors":"Na Li ,&nbsp;Xuemei Li ,&nbsp;Song Ding","doi":"10.1016/j.engappai.2026.113984","DOIUrl":"10.1016/j.engappai.2026.113984","url":null,"abstract":"<div><div>The sustainable development of the marine economy is inseparable from green development. However, the spatial correlations and dynamic evolution of green development present significant forecasting challenges. To address this, this paper first constructs an evaluation index system for the green development of marine economy based on the Driver-Pressure-State-Impact-Response (<em>DPSIR</em>) framework. Furthermore, the grey Bernoulli model is improved in both spatial and temporal dimensions for forecasting purposes. Interaction terms between spatial distance matrices and variables are introduced to capture spatial correlations, while a time-varying component is incorporated to reflect dynamic evolution. These enhancements enable the model to more effectively characterize the spatial, temporal, and nonlinear features of the green development of marine economy. Additionally, the model’s hyperparameters and weighting coefficients are optimized using the whale optimization algorithm. For validation, an empirical study is conducted across China’s 11 coastal provinces and municipalities. Systematic analyses show that the proposed model has high predictive accuracy. Robustness tests and sensitivity analysis further confirm that the model demonstrates excellent stability, reliability, and generalization capability.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"170 ","pages":"Article 113984"},"PeriodicalIF":8.0,"publicationDate":"2026-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146193287","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}