Advanced Engineering Informatics最新文献

{"title":"Turning crisis into strength: development of a comprehensive antifragility framework for safety processes in industries","authors":"Mojtaba Emkani ,&nbsp;Moslem Alimohammadlou ,&nbsp;Rosanna Cousins ,&nbsp;Esmaeil Zarei ,&nbsp;Ramezan Khosravi ,&nbsp;Mojtaba Kamalinia ,&nbsp;Mehdi Jahangiri","doi":"10.1016/j.aei.2026.104336","DOIUrl":"10.1016/j.aei.2026.104336","url":null,"abstract":"<div><div>Conventional resilience approaches that simply aim to return functions to their pre-crisis state are insufficient for effective management in today’s industrial sector. This article introduces a comprehensive and innovative framework for antifragility that can use crises as opportunities to put better safety processes in place in industries. The framework was designed drawing on data from the scientific literature in vulnerability, resilience and antifragility, and the collective wisdom of experts in the field. Criteria from this qualitative data were then prioritized using the Spherical Fuzzy Delphi (SF-Delphi) method, and finally, with the participation of expert panels, the Comprehensive Antifragility Framework was developed. The framework is structured around three pillars: <em>Direction</em> (responsible leadership, systematic environmental awareness, aligned policymaking), <em>Execution</em> (criteria for achieving antifragility in the form of human resources, organizational structure, capital and resources, data and knowledge management, technology, risk and crisis management, and organizational safety and security), and <em>Results</em> (the organization’s recoverability, learning, reliability, flexibility and dynamism). The proposed framework represents a comprehensive, evidence-based antifragility framework for industries seeking a secure, adaptive, and forward-thinking future, that can make them resilient to system safety crises, and able to thrive in the face of disruption.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"71 ","pages":"Article 104336"},"PeriodicalIF":9.9,"publicationDate":"2026-01-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146023034","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Dynamic reliability analysis for complex multi-state systems of more-electric aircraft: A lightweight DBN method based on E-TrMF and interval grey number","authors":"Jiayu Chen ,&nbsp;Xuhang Wang ,&nbsp;Qinhua Lu ,&nbsp;Min Xie ,&nbsp;Hongjuan Ge ,&nbsp;Dong Zhou","doi":"10.1016/j.aei.2026.104345","DOIUrl":"10.1016/j.aei.2026.104345","url":null,"abstract":"<div><div>Bayesian network-based reliability analysis approaches can effectively address structural limitations inherent in complex multi-state systems. However, three challenges of high computational complexity, multiple failure states, and multi-source uncertainty in the more-electric aircraft system pose significant obstacles to achieve accurate reliability analysis. To bridge these gaps, this study proposes a dynamic reliability analysis method for complex multi-state systems of more-electric aircraft based on fuzzy support radius and Markov chain enhanced trapezoidal membership function (E-TrMF) and interval grey number-lightweight dynamic Bayesian network (EI-LDBN). First, a grey Bayesian information criterion-based dynamic Bayesian network lightweighting strategy is developed to eliminate redundant structures of the DBN in the complex multi-state systems and enhance the model efficiency, which constructs a LDBN architecture. Then, the E-TrMF is developed by an integrating trapezoidal membership function with the fuzzy support radius and a discrete-time Markov chain, which dynamically quantifies multi-failure states of key components in the multi-state systems. Subsequently, a grey conditional probability table is developed for the LDBN model by using interval grey numbers to enable failure analysis under dynamic conditions, which effectively addresses multi-source uncertainty. Finally, an EI-LDBN framework is proposed to achieve accurate and dynamic reliability analysis for the complex multi-state systems of more-electric aircraft. A case of the starter-generator system (SGS) of more-electric aircraft has been conducted, and the experiment results demonstrate that the proposed EI-LDBN method can effectively achieve the lightweight modeling and dynamic reliability analysis with 39.1% and 63% efficiency improvements compared with EI-DBN and DFT methods under multi-source uncertainty.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"71 ","pages":"Article 104345"},"PeriodicalIF":9.9,"publicationDate":"2026-01-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146023037","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Dynamic sequential and non-sequential learning for predicting diaphragm wall deflection and ground subsidence in deep excavation of building construction","authors":"Min-Yuan Cheng,&nbsp;Akhmad F.K. Khitam,&nbsp;Quoc-Tuan Vu,&nbsp;Yi-Fan Ying","doi":"10.1016/j.aei.2026.104356","DOIUrl":"10.1016/j.aei.2026.104356","url":null,"abstract":"<div><div>With the expansion of subway infrastructures across urban China, deep excavations have become increasingly common. Predicting deformation during excavation is crucial during the construction of Mass Rapid Transportation (MRT) station buildings, as excessive displacement may result in severe economic and safety consequences. Traditional numerical methods and existing AI models fail to simultaneously process time-dependent and time-independent factors, limiting their predictive accuracy and adaptability. To address this problem, this study proposes a novel Artificial Satellite Search Algorithm-Neural Network-Revolving Gate Fourier Transformation (ASSA-NN-RGFT) framework with three key innovations: (1) a dual-stream architecture that separately processes sequential data through RGFT’s frequency-domain analysis and non-sequential data through NN; (2) a revolving mechanism that performs iterative FFT-iFFT transformations with cross-attention, capturing periodic deformation patterns missed by conventional recurrent models; and (3) automated ASSA-based hyperparameter optimization that eliminates the need for manual calibration. This model, trained to predict diaphragm wall deflection and ground settlement, is a tool tailored for real-time deformation monitoring and early warning. In comparison against six advanced AI models, the proposed ASSA-NN-RGFT achieved the best overall performance, earning the lowest error (wall displacement: MAPE = 3.54 %; ground settlement: MAPE = 5.78 %) and highest correlation metric (wall displacement: R² = 0.952; ground settlement: R² = 0.842). Real-world validation across ten MRT projects achieved 85.42 % high-accuracy predictions for wall displacement and 5.89 % average error for ground settlement. The exceptional predictive performance, dynamic adaptability, and strong reliability demonstrated by the proposed ASSA-NN-RGFT model support its viability as a decision-support system in excavation risk management.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"71 ","pages":"Article 104356"},"PeriodicalIF":9.9,"publicationDate":"2026-01-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146023036","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A multi-module optimized UAV-YOLOv11 model and 3D coordinate reconstruction method for UAV-based monitoring of cable dome nodes","authors":"Lingqian Wen ,&nbsp;Hui Lv ,&nbsp;Xin Xie ,&nbsp;Xiaodong Feng ,&nbsp;Yuanqi Li ,&nbsp;Jiayang Lv ,&nbsp;Feifei Liu","doi":"10.1016/j.aei.2026.104357","DOIUrl":"10.1016/j.aei.2026.104357","url":null,"abstract":"<div><div>Accurate monitoring of the spatial position of cable dome structure nodes is crucial for their structural safety. Traditional measurement methods are inefficient, while existing visual techniques are limited by challenges such as the small scale of node targets, strong background interference, and insufficient localization accuracy. To address this, this paper proposes a knowledge-guided perception-solution collaborative framework, achieving automated reconstruction from UAV images to 3D spatial coordinates. This framework first constructs a multi-module optimized UAV-YOLOv11 detection model. By integrating Bi-level Routing Attention, lightweight convolution, and Normalized Wasserstein Distance loss, the detection accuracy [email protected]:0.95 is improved to 88.7%. In the reconstruction stage, by introducing geometric, topological, and symmetry constraints derived from the MATLAB form-finding algorithm, Python is guided to achieve coordinate back-projection and optimization solution, forming an inverse reconstruction mechanism, ultimately achieving an average point measurement accuracy of 2.72 cm. This research provides a reliable technical path for the construction and health monitoring of large-span spatial structures, promoting the practical application of knowledge-embedded engineering informatics methods in structural monitoring.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"71 ","pages":"Article 104357"},"PeriodicalIF":9.9,"publicationDate":"2026-01-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146022936","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A novel multi-task sequential network integrating wear information for tool breakage monitoring","authors":"Shenping Mei ,&nbsp;Xuandong Mo ,&nbsp;Mingyuan Xia ,&nbsp;Xiaofeng Hu","doi":"10.1016/j.aei.2026.104342","DOIUrl":"10.1016/j.aei.2026.104342","url":null,"abstract":"<div><div>Tool breakage monitoring (TBM) during machining processes is crucial for ensuring manufacturing safety and product quality. Although data-driven methods have made significant progress in the field of TBM, existing studies generally overlook the intrinsic correlation between tool breakage and wear processes, as the probability and severity of tool breakage vary with wear accumulation. Meanwhile, conservative process parameters in industrial practices lead to a scarcity of breakage samples, and the normal samples caused by fluctuations in working conditions exhibit multiple pattern distributions, which further increase monitoring difficulties. To address these issues, this paper proposes a novel multi-task sequential network that integrates wear information to achieve tool breakage monitoring. Unlike traditional multi-task learning (MTL), this method jointly learns tool wear prediction and breakage monitoring tasks in a sequential manner. The two tasks share a feature extraction module, and the predicted continuous wear value is fed as conditional information into the subsequent breakage monitoring module, enabling the monitoring model to adapt to different wear states. Additionally, the weighted multi-scale distance (WMSD) loss function designed for the breakage monitoring module integrates Euclidean distance, angular difference, and Latent feature distance, enhancing the capability to model the distribution pattern of normal samples and strengthening its robustness. Experimental findings indicate that the proposed method achieved favorable results on both the actual machining dataset of aerospace engine casings and the PHM2010 dataset, validating its effectiveness and broad applicability in practical scenarios.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"71 ","pages":"Article 104342"},"PeriodicalIF":9.9,"publicationDate":"2026-01-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146023038","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"AI-driven intelligent hazard monitoring for major petroleum projects under the Belt and Road Initiative: construction of an indicator system based on DPGT and RAG-HLLM","authors":"Ke He,&nbsp;Changfeng Wang","doi":"10.1016/j.aei.2026.104340","DOIUrl":"10.1016/j.aei.2026.104340","url":null,"abstract":"<div><div>The major petroleum projects under the Belt and Road Initiative (BRI) encounter intricate hazard problems. Establishing a comprehensive and scientific hazard monitoring indicator system is important to realize early identification and proactive warning of risks. Therefore, this study innovatively proposes the “Dual-Path Grounded Theory” (DPGT) as a new paradigm for human–machine collaborative qualitative research. The paradigm combines the depth of human analysis with the broadness of LLMs’ processing through a system that allows for parallel coding, cross-validation, and complementary integration between human and LLM paths. It overcomes the limitations of depending merely on ineffective manual analysis or unreliable fully automatic methods. In implementation, the human-driven path utilizes the KeyBERT model and K-means clustering algorithm to help complete the three stages of coding. The LLM-driven path assesses the performance of DeepSeek-R1, Llama3, and Qwen3 models at various coding stages and chooses DeepSeek-R1 as the main engine. This is combined with Retrieval-Augmented Generation (RAG) to build a hybrid LLM deployment architecture (RAG-HLLM), making full use of superior cloud-based reasoning abilities and protecting local data privacy. According to the DPGT framework, after several rounds of integration and screening, a hazard monitoring indicator system including 2 major categories, 8 first-level indicators, 27 second-level indicators, and 109 specific factors was finally established. Validation indicates that this system has passed the theoretical saturation test and has achieved a recall rate of 92.61% on the International Association of Oil and Gas Producers (IOGP) standard database, which greatly surpasses the results obtained from single-path methods. Methodologically, this study pioneers a new paradigm for human–machine collaborative qualitative research. Technologically, the proposed hybrid architecture balances privacy concerns with processing efficiency. Practically, it provides a dependable tool for safety management in major energy projects, exhibiting significant theoretical value and broad engineering applicability.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"71 ","pages":"Article 104340"},"PeriodicalIF":9.9,"publicationDate":"2026-01-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146022933","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Physics-informed embodied intelligence in the foundation model era: Advancing robot manipulation for smart manufacturing","authors":"Shufei Li ,&nbsp;Cheng Liu ,&nbsp;Jianzhuang Zhao ,&nbsp;Pai Zheng ,&nbsp;Xi Vincent Wang ,&nbsp;Lihui Wang","doi":"10.1016/j.aei.2026.104370","DOIUrl":"10.1016/j.aei.2026.104370","url":null,"abstract":"<div><div>In recent years, foundation models have emerged as a transformative force in artificial intelligence, enabling a paradigm shift in how robotic intelligence is conceived and implemented. However, despite rapid progress, the integration of physical reasoning into these large-scale models remains limited, leaving a gap between intricate manipulation tasks to complex manufacturing processes. Aligned with the current trends in robotic cognition development, we provide a comprehensive review of the theoretical foundations that underpin physics-informed approaches and analyze the architectural frameworks that merge data-driven learning with physical laws and constraints. By synthesizing the latest developments in both academic research and industrial applications, this paper highlights how the fusion of these domains not only enhances the robustness and interpretability of robotic systems but also accelerates their transition from simulation to real-world deployment. Furthermore, we discuss key case studies, benchmark techniques, and emerging directions, offering insights into the synergies and trade-offs inherent in current methodologies. Our analysis outlines promising avenues for future research, aiming to further bridge the gap between theoretical advancements and practical implementations in robotics. This survey positions physics-informed embodied intelligence as a critical enabler for smart manufacturing in the foundation model era.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"71 ","pages":"Article 104370"},"PeriodicalIF":9.9,"publicationDate":"2026-01-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146023022","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"MSFRNet: A machining similarity feature recognition network based on a Kolmogorov–Arnold enhanced graph neural mixer","authors":"Yun Ren ,&nbsp;Xiaoqing Tian ,&nbsp;Jiang Han ,&nbsp;Lian Xia ,&nbsp;Chao Shi ,&nbsp;Jiacheng Lin ,&nbsp;Yu Huang ,&nbsp;Xiong Peng","doi":"10.1016/j.aei.2026.104365","DOIUrl":"10.1016/j.aei.2026.104365","url":null,"abstract":"<div><div>Machining feature recognition plays a crucial role in process planning, as it assists in process arrangement by extracting dimensional and geometric information, and serves as a key step toward automatic programming. The recognition accuracy directly affects the quality of tool paths, Especially in the discrimination of similar machining features, features such as holes and shafts often exhibit a high degree of similarity in both geometric appearance and topological structure. Existing deep learning methods struggle to capture the underlying topological relationships among machining features and therefore tend to misclassify highly similar features as the same category. This misidentification adversely affects subsequent process planning and can significantly degrade machining quality and overall manufacturing efficiency. To address this issue, we propose the Machining Similar Feature Recognition Network based on a Kolmogorov–Arnold-Augmented Graph Neural Mixer (MSFRNet). Among these components, Kolmogorov–Arnold Network(KAN) provides learnable spline functions capable of accurately fitting arbitrarily complex geometric or topological mappings, making it more sensitive to intricate geometric relationships. Based on this advantage,KAN is incorporated into the local feature learning stage of the Graph Neural Network(GNN) to enhance the model’s capacity for fine-grained local geometric approximation. Optimizing the local and global feature vectors in GNN using KAN. Ultimately enhancing the recognition capability of MSFRNet for typical machining features. Meanwhile, the concepts of convexity,concavity and openness are introduced to further improve the discrimination of similar machining features. Convex and concave connections reflect the topological relationships between adjacent surfaces and serve as reliable indicators for distinguishing features with highly similar shapes. In addition, openness enables the detection of whether the topological boundary of a machining feature is open or closed, providing an additional criterion for differentiating certain types of similar machining features. The purpose is to leverage these two criteria to enhance the recognition capability of MSFRNet for similar machining features. Extensive experiments on publicly available benchmark datasets demonstrate that MSFRNet achieves outstanding performance in machining feature recognition tasks, reaching an accuracy of 99.95%. In the task of recognizing similar machining features, the method achieves an accuracy of 98.6%.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"71 ","pages":"Article 104365"},"PeriodicalIF":9.9,"publicationDate":"2026-01-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146022932","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Causal feature selection framework for stable soft sensor modeling based on time-delayed cross mapping","authors":"Shi-Shun Chen ,&nbsp;Xiao-Yang Li ,&nbsp;Enrico Zio","doi":"10.1016/j.aei.2026.104337","DOIUrl":"10.1016/j.aei.2026.104337","url":null,"abstract":"<div><div>Soft sensor modeling plays a crucial role in process monitoring. Causal feature selection can enhance the performance of soft sensor models in industrial applications. However, existing methods ignore two critical characteristics of industrial processes. Firstly, causal relationships between variables always involve time delays, whereas most causal feature selection methods investigate causal relationships in the same time dimension. Secondly, variables in industrial processes are often interdependent, which contradicts the decorrelation assumption of traditional causal inference methods. Consequently, soft sensor models based on existing causal feature selection approaches often lack sufficient accuracy and stability. To overcome these challenges, this paper proposes a causal feature selection framework based on time-delayed cross mapping. Time-delayed cross mapping employs state space reconstruction to effectively handle interdependent variables in causality analysis, and considers varying causal strength across time delay. Time-delayed convergent cross mapping (TDCCM) is introduced for total causal inference, and time-delayed partial cross mapping (TDPCM) is developed for direct causal inference. Then, in order to achieve automatic feature selection, an objective feature selection strategy is presented. The causal threshold is automatically determined based on the model performance on the validation set, and the causal features are then selected. Two real-world case studies show that TDCCM achieves the highest average performance, while TDPCM improves soft sensor stability and performance in the worst scenario. On average over the two cases, TDCCM decreases root mean square error (RMSE) by about 8.93% compared with the best existing feature selection method, and TDPCM further decreases RMSE in the worst scenario by about 7.69% relative to TDCCM. The code is publicly available at <span><span>https://github.com/dirge1/TDPCM</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"71 ","pages":"Article 104337"},"PeriodicalIF":9.9,"publicationDate":"2026-01-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145977713","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Task scheduling of many-objective industrial workflow applications via co-evolutionary swarm optimizer with learnable offspring generators","authors":"Yongxiang Li ,&nbsp;Jiajun Zhou ,&nbsp;Chao Lu ,&nbsp;Liang Gao","doi":"10.1016/j.aei.2026.104325","DOIUrl":"10.1016/j.aei.2026.104325","url":null,"abstract":"<div><div>Nowadays, the proliferation of industrial internet of things brings the skyrocketing rise in large scale data analysis and processing. As the base to realize industrial intelligence, complex industrial applications are usually accompanied by computational workflows, which typically run in heterogeneous computing resources on cloud, and many factors such as makespan, cost, reliability, energy consumption and load balancing need to be optimized simultaneously. As an NP-hard problem, how to determine proper computing nodes for each task of workflow application with many tightly coupled performance concerns is extraordinarily challenging. Focusing on this many-objective optimization problem, we present a novel learnable co-evolutionary scheduler where multiple learnable offspring generators are integrated and work cooperatively toward robust search capability, the limited computational budgets are invested into multiple generators automatically by learning from historical successful experience and exploiting heuristic information. In addition, an adaptive cluster based ranking mechanism is devised to preserve prominent solutions in environmental selection of many-objective space, which is expected to leverage the solution diversity and expedite the convergence. Empirical studies on real-life workflows and extensive synthetic benchmark test suites confirm that our proposal outperforms or is at least comparable to other state-of-the-art contenders.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"71 ","pages":"Article 104325"},"PeriodicalIF":9.9,"publicationDate":"2026-01-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145977714","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}