Pub Date : 2026-01-01DOI: 10.1016/j.aej.2025.12.041
Chaudry Masood Khalique , Oke Davies Adeyemo
This paper significantly provides various analytical examinations of the first extended Vakhnenko–Parkes equation, which has never been explored before. To this model, abundant structures of various exact periodic solutions to this equation through the use of the extended Jacobi elliptic function expansion technique is abundantly computed for the first time. This approach employs three different auxiliary nonlinear differential equations to generate the solutions. Consequently, various cnoidal, snoidal, dnoidal, and complex snoidal wave solutions of notes were obtained. We further explore the wave dynamics of these periodic solutions in three and two dimensions using computer software. Moreover, Lie group analysis is utilized to generate the symmetries of the equation. Thereafter, conserved vectors associated with the equation are constructed through the application of Ibragimov’s conserved theorem using the formal Lagrangian of the model. The results, whose significance is demonstrated in physical sciences and technology in this research, can be very useful for researchers in relevant fields for further analysis and applications. All these results are new and unique, and they complement the work previously done on the model, thus underscoring the originality of this research.
{"title":"Cnoidal, snoidal, dnoidal wave solutions of the first extended 3D Vakhnenko–Parkes equation together with its conservation laws and various life applications","authors":"Chaudry Masood Khalique , Oke Davies Adeyemo","doi":"10.1016/j.aej.2025.12.041","DOIUrl":"10.1016/j.aej.2025.12.041","url":null,"abstract":"<div><div>This paper significantly provides various analytical examinations of the first extended Vakhnenko–Parkes equation, which has never been explored before. To this model, abundant structures of various exact periodic solutions to this equation through the use of the extended Jacobi elliptic function expansion technique is abundantly computed for the first time. This approach employs three different auxiliary nonlinear differential equations to generate the solutions. Consequently, various cnoidal, snoidal, dnoidal, and complex snoidal wave solutions of notes were obtained. We further explore the wave dynamics of these periodic solutions in three and two dimensions using computer software. Moreover, Lie group analysis is utilized to generate the symmetries of the equation. Thereafter, conserved vectors associated with the equation are constructed through the application of Ibragimov’s conserved theorem using the formal Lagrangian of the model. The results, whose significance is demonstrated in physical sciences and technology in this research, can be very useful for researchers in relevant fields for further analysis and applications. All these results are new and unique, and they complement the work previously done on the model, thus underscoring the originality of this research.</div></div>","PeriodicalId":7484,"journal":{"name":"alexandria engineering journal","volume":"135 ","pages":"Pages 20-35"},"PeriodicalIF":6.8,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145897859","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-01DOI: 10.1016/j.aej.2025.12.065
Adel Alshamrani , Ahmed M. Alghamdi
Traditional intrusion detection systems require extensive training data for each attack type, creating vulnerabilities in UAV networks where novel threats frequently emerge. We present ZeroUAV, a zero-shot attack detection framework using foundation models to identify unseen attack patterns without labeled examples. Our transformer-based architecture employs cross-modal attention to align network traffic with semantic attack descriptions, enabling classification based on conceptual similarity. The framework incorporates attack ontology, contrastive learning, and meta-learning for rapid adaptation. Evaluation on UAV-NIDD dataset shows 87.3 % zero-shot accuracy and 94.6 % accuracy with five examples per attack type, significantly outperforming supervised methods. The system achieves under 10 ms inference latency for real-time UAV deployment. Our contributions include the first foundation model for UAV cybersecurity, a novel zero-shot learning framework, and validation demonstrating practical viability for evolving threat landscapes.
{"title":"Zero-shot attack detection in UAV networks using foundation models","authors":"Adel Alshamrani , Ahmed M. Alghamdi","doi":"10.1016/j.aej.2025.12.065","DOIUrl":"10.1016/j.aej.2025.12.065","url":null,"abstract":"<div><div>Traditional intrusion detection systems require extensive training data for each attack type, creating vulnerabilities in UAV networks where novel threats frequently emerge. We present ZeroUAV, a zero-shot attack detection framework using foundation models to identify unseen attack patterns without labeled examples. Our transformer-based architecture employs cross-modal attention to align network traffic with semantic attack descriptions, enabling classification based on conceptual similarity. The framework incorporates attack ontology, contrastive learning, and meta-learning for rapid adaptation. Evaluation on UAV-NIDD dataset shows 87.3 % zero-shot accuracy and 94.6 % accuracy with five examples per attack type, significantly outperforming supervised methods. The system achieves under 10 ms inference latency for real-time UAV deployment. Our contributions include the first foundation model for UAV cybersecurity, a novel zero-shot learning framework, and validation demonstrating practical viability for evolving threat landscapes.</div></div>","PeriodicalId":7484,"journal":{"name":"alexandria engineering journal","volume":"136 ","pages":"Pages 105-124"},"PeriodicalIF":6.8,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145975061","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-01DOI: 10.1016/j.aej.2026.01.017
Mingyue Wei, Weidong Zhao, Ning Jia, Xianhui Liu, Zhen Xu
Lead frames are key components in semiconductor packaging, where undetected surface defects may lead to reliability issues and economic losses. Existing surface defect detection methods are limited by large variations in defect scale, resulting in reduced accuracy and continued dependence on manual inspection. To address these challenges, a lightweight multi-scale defect detection network (MSDD-Net) is proposed for automated lead frame inspection. The network employs a dual-branch dynamic hierarchical fusion backbone (DDHFNet) to extract discriminative multi-scale texture features. A cross-scale feature amalgamation module (GSMD-CFA) is introduced, which integrates gated sampling with multi-dilation convolution to enhance both long-range and short-range feature interactions. A customized loss function is further designed to improve robustness by emphasizing hard samples during training. The proposed network contains 19.68 million parameters, meeting the deployment constraints of industrial real-time inspection while preserving high detection accuracy. A lead frame surface defect dataset comprising 43 defect categories is constructed to reflect real industrial conditions. On this dataset, the proposed method achieves a mean Average Precision (mAP) of 86.8%, with performance improvements of 6.7, 4.0, and 0.9 percentage points for small, medium, and large defects, respectively, compared with the baseline. Experiments on a public steel surface defect dataset further demonstrate the generalization capability of the proposed method.
{"title":"MSDD-Net: A lightweight multi-scale defect detection network for industrial lead frame inspection","authors":"Mingyue Wei, Weidong Zhao, Ning Jia, Xianhui Liu, Zhen Xu","doi":"10.1016/j.aej.2026.01.017","DOIUrl":"10.1016/j.aej.2026.01.017","url":null,"abstract":"<div><div>Lead frames are key components in semiconductor packaging, where undetected surface defects may lead to reliability issues and economic losses. Existing surface defect detection methods are limited by large variations in defect scale, resulting in reduced accuracy and continued dependence on manual inspection. To address these challenges, a lightweight multi-scale defect detection network (MSDD-Net) is proposed for automated lead frame inspection. The network employs a dual-branch dynamic hierarchical fusion backbone (DDHFNet) to extract discriminative multi-scale texture features. A cross-scale feature amalgamation module (GSMD-CFA) is introduced, which integrates gated sampling with multi-dilation convolution to enhance both long-range and short-range feature interactions. A customized loss function is further designed to improve robustness by emphasizing hard samples during training. The proposed network contains 19.68 million parameters, meeting the deployment constraints of industrial real-time inspection while preserving high detection accuracy. A lead frame surface defect dataset comprising 43 defect categories is constructed to reflect real industrial conditions. On this dataset, the proposed method achieves a mean Average Precision (mAP) of 86.8%, with performance improvements of 6.7, 4.0, and 0.9 percentage points for small, medium, and large defects, respectively, compared with the baseline. Experiments on a public steel surface defect dataset further demonstrate the generalization capability of the proposed method.</div></div>","PeriodicalId":7484,"journal":{"name":"alexandria engineering journal","volume":"136 ","pages":"Pages 45-61"},"PeriodicalIF":6.8,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145975065","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-01DOI: 10.1016/j.aej.2026.01.003
Wessam M. Salama , Moustafa H. Aly
This paper introduces a new hybrid DL framework, CVAE-UNETR-ResNet50-VGG16, for accurate brain tumor segmentation and classification from MRI and BraTS2021 scans. The proposed model integrates a Convolutional Variational Autoencoder (CVAE) for synthetic MRI and BraTS2021 data generation, a UNET Transformer (UNETR) for enhanced spatial segmentation through global self-attention, and ResNet50 and VGG16 networks for robust multi-scale feature classification. Moreover, data augmentation technique is proposed based on both CVAE and diffusion technique. Experimental results on the BraTS2021 dataset demonstrate a 2.57 % overall improvement in segmentation and classification performance compared to conventional UNET-based approaches. The model achieved a Dice Similarity Coefficient (DSC) of 97.45 %, Intersection over Union (IoU) of 95.67 %, and a classification accuracy of 99.35 %, representing a 3.1 % reduction in segmentation error and a 2.4 % increase in classification accuracy over benchmark models. The inference time per image is 1.6541 s on a system with 13 GB RAM, confirming its computational efficiency for clinical deployment. By effectively combining generative modeling, transformer-based segmentation, and deep feature classification, the proposed CVAE-UNETR-ResNet50-VGG16 framework establishes a new performance benchmark for automated brain tumor analysis, offering a quantifiable step forward, 2.5–3 % improvement, in diagnostic precision, computational efficiency, and medical imaging reliability. Thus, the CVAE-UNETR-ResNet50-VGG16 model offers a measurable 2–3 % performance improvement over current techniques, creating a basis for AI-assisted brain tumor diagnosis and treatment planning. This advancement supports the broader goal of AI-driven healthcare, enhancing early diagnosis and treatment planning for neurological disorders. This hybrid design bridges the gap between data-driven inference and structural MRI priors, enhancing both interpretability and precision in clinical decision-making.
{"title":"Brain tumor segmentation and classification: A CVAE-UNETR-ResNet50-VGG16 hybrid deep learning approach","authors":"Wessam M. Salama , Moustafa H. Aly","doi":"10.1016/j.aej.2026.01.003","DOIUrl":"10.1016/j.aej.2026.01.003","url":null,"abstract":"<div><div>This paper introduces a new hybrid DL framework, CVAE-UNETR-ResNet50-VGG16, for accurate brain tumor segmentation and classification from MRI and BraTS2021 scans. The proposed model integrates a Convolutional Variational Autoencoder (CVAE) for synthetic MRI and BraTS2021 data generation, a UNET Transformer (UNETR) for enhanced spatial segmentation through global self-attention, and ResNet50 and VGG16 networks for robust multi-scale feature classification. Moreover, data augmentation technique is proposed based on both CVAE and diffusion technique. Experimental results on the BraTS2021 dataset demonstrate a 2.57 % overall improvement in segmentation and classification performance compared to conventional UNET-based approaches. The model achieved a Dice Similarity Coefficient (DSC) of 97.45 %, Intersection over Union (IoU) of 95.67 %, and a classification accuracy of 99.35 %, representing a 3.1 % reduction in segmentation error and a 2.4 % increase in classification accuracy over benchmark models. The inference time per image is 1.6541 s on a system with 13 GB RAM, confirming its computational efficiency for clinical deployment. By effectively combining generative modeling, transformer-based segmentation, and deep feature classification, the proposed CVAE-UNETR-ResNet50-VGG16 framework establishes a new performance benchmark for automated brain tumor analysis, offering a quantifiable step forward, <span><math><mo>≈</mo></math></span>2.5–3 % improvement, in diagnostic precision, computational efficiency, and medical imaging reliability. Thus, the CVAE-UNETR-ResNet50-VGG16 model offers a measurable 2–3 % performance improvement over current techniques, creating a basis for AI-assisted brain tumor diagnosis and treatment planning. This advancement supports the broader goal of AI-driven healthcare, enhancing early diagnosis and treatment planning for neurological disorders. This hybrid design bridges the gap between data-driven inference and structural MRI priors, enhancing both interpretability and precision in clinical decision-making.</div></div>","PeriodicalId":7484,"journal":{"name":"alexandria engineering journal","volume":"135 ","pages":"Pages 433-449"},"PeriodicalIF":6.8,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145923558","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-01DOI: 10.1016/j.aej.2025.12.032
Ahmed M. Alghamdi , Adel Bahaddad , Khalid Almarhabi , Asmaa A. Al-Zobidi
Hajj and Umrah services annually attract millions of pilgrims to Saudi Arabia, making their efficient management crucial to achieving Vision 2030’s objectives. This paper explores the use of artificial intelligence (AI) and machine learning to predict and optimize key performance indicators for these services. We propose an AI-driven framework that processes vast datasets to enhance decision making, improve service provision, and optimize the pilgrimage experience. Our results demonstrate significant improvements in KPI prediction accuracy, supporting Saudi Arabia’s efforts to advance the quality of Hajj and Umrah services while aligning with Vision 2030’s goals.
{"title":"Predictive analysis of Hajj and Umrah performance using key performance indicators (KPIs) and machine learning (ML)","authors":"Ahmed M. Alghamdi , Adel Bahaddad , Khalid Almarhabi , Asmaa A. Al-Zobidi","doi":"10.1016/j.aej.2025.12.032","DOIUrl":"10.1016/j.aej.2025.12.032","url":null,"abstract":"<div><div>Hajj and Umrah services annually attract millions of pilgrims to Saudi Arabia, making their efficient management crucial to achieving Vision 2030’s objectives. This paper explores the use of artificial intelligence (AI) and machine learning to predict and optimize key performance indicators for these services. We propose an AI-driven framework that processes vast datasets to enhance decision making, improve service provision, and optimize the pilgrimage experience. Our results demonstrate significant improvements in KPI prediction accuracy, supporting Saudi Arabia’s efforts to advance the quality of Hajj and Umrah services while aligning with Vision 2030’s goals.</div></div>","PeriodicalId":7484,"journal":{"name":"alexandria engineering journal","volume":"135 ","pages":"Pages 400-406"},"PeriodicalIF":6.8,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145923613","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-01DOI: 10.1016/j.aej.2025.11.049
Jih-Hsing Chang , S. Selvaraj , S. Manikandan , S. Nagarani , Saiful Islam , Mohd Shkir , Arun Kumar Senthilkumar , Ashwini J. John , Selvarajan Ethiraj , Sambasivam Sangaraju , Mohanraj Kumar
Carbon composites made from fly ash represent a significant innovation in environmentally friendly construction materials. These composites offer multiple advantages, including CO2 capture and improved energy efficiency, which leads to a circular economy. Fly ash, which is a by-product of coal combustion, has been integrated into carbon-based materials to create composites that enhance concrete's structural performance while also supporting environmental sustainability. Given that a high surface area and porosity are crucial for CO2 adsorption, these composites are expected to serve as an environmentally friendly alternative for producing construction materials with minimal greenhouse gas emissions. Carbon materials intermixed in fly ash composites increase concrete mechanical strength and service life, help construction projects last longer, and extend their useful lives more cost-effectively. Besides incorporating QWR into composites, this approach reduces the energy required for processing since, unlike conventional materials, they are manufactured at lower temperatures and with less energy-demanding processes. CO2 adsorption capacity of fly ash carbon composites can be increased due to the synergistic effects between carbon and fly ash via higher CO2 capture efficiency by research. Moreover, the use of industrial waste fly ash is in line with the principles of circular economy, lessening waste and promoting resource efficiency. The fly ash carbon composites could enhance the performance of construction materials, at the same time, it can solve problems related to CO2 emission and energy production. Adoption of the same by the construction industry can be useful in achieving sustainability goals SDG11 and reducing the carbon footprint (SDG13) of infrastructure projects to attract greener, more energy-efficient building practices.
{"title":"Fly ash carbon composites: A breakthrough in CO2 capture and energy efficiency","authors":"Jih-Hsing Chang , S. Selvaraj , S. Manikandan , S. Nagarani , Saiful Islam , Mohd Shkir , Arun Kumar Senthilkumar , Ashwini J. John , Selvarajan Ethiraj , Sambasivam Sangaraju , Mohanraj Kumar","doi":"10.1016/j.aej.2025.11.049","DOIUrl":"10.1016/j.aej.2025.11.049","url":null,"abstract":"<div><div>Carbon composites made from fly ash represent a significant innovation in environmentally friendly construction materials. These composites offer multiple advantages, including CO<sub>2</sub> capture and improved energy efficiency, which leads to a circular economy. Fly ash, which is a by-product of coal combustion, has been integrated into carbon-based materials to create composites that enhance concrete's structural performance while also supporting environmental sustainability. Given that a high surface area and porosity are crucial for CO<sub>2</sub> adsorption, these composites are expected to serve as an environmentally friendly alternative for producing construction materials with minimal greenhouse gas emissions. Carbon materials intermixed in fly ash composites increase concrete mechanical strength and service life, help construction projects last longer, and extend their useful lives more cost-effectively. Besides incorporating QWR into composites, this approach reduces the energy required for processing since, unlike conventional materials, they are manufactured at lower temperatures and with less energy-demanding processes. CO<sub>2</sub> adsorption capacity of fly ash carbon composites can be increased due to the synergistic effects between carbon and fly ash via higher CO<sub>2</sub> capture efficiency by research. Moreover, the use of industrial waste fly ash is in line with the principles of circular economy, lessening waste and promoting resource efficiency. The fly ash carbon composites could enhance the performance of construction materials, at the same time, it can solve problems related to CO<sub>2</sub> emission and energy production. Adoption of the same by the construction industry can be useful in achieving sustainability goals SDG11 and reducing the carbon footprint (SDG13) of infrastructure projects to attract greener, more energy-efficient building practices.</div></div>","PeriodicalId":7484,"journal":{"name":"alexandria engineering journal","volume":"135 ","pages":"Pages 450-473"},"PeriodicalIF":6.8,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145974097","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-01DOI: 10.1016/j.aej.2025.12.069
Wei Liu, Dexian Li
Mountain flood disasters in rugged terrains pose significant challenges due to rapid onset, complex spatiotemporal dynamics, and data scarcity, where traditional hydrological models and pairwise graph neural networks struggle to capture multi-scale dependencies and uncertainty in propagation patterns. This study proposes the Adaptive Spatiotemporal Uncertainty-Guided Neural Network (ASTUNN), a hybrid framework that synergistically combines Bidirectional Gated Recurrent Units (BiGRU) for temporal modeling, Spherical Manifold Graph Learning (SMGL) for non-Euclidean spatial analysis, Fractional-Order Dynamic Attention (FODA) for long-memory patterns, Stochastic Variational Inference (SVI) for uncertainty quantification, and Adaptive Feature Synthesis (AFS) for multi-scale fusion. Key innovations include: (1) hyperedge-aware spatiotemporal message passing with fractional-order attention to model higher-order interactions and long-range dependencies in river networks and terrain gradients; and (2) stochastic variational uncertainty estimation to provide calibrated probabilistic forecasts and prevention capability rankings. These contributions overcome limitations of static graphs and deterministic models under rapid environmental changes. Validated on multi-source hydrological datasets from seven high-risk mountainous regions in southwest China, ASTUNN achieves an AUC-ROC of 0.947, MAE of 0.103 for prevention capability rankings, and ECE of 0.029, outperforming state-of-the-art baselines by 15–25 % while reducing false alarms by 18 % and enabling early warnings up to 48 h ahead.
{"title":"ASTUNN: An enhanced spatiotemporal uncertainty guided neural network for flood management in mountainous areas","authors":"Wei Liu, Dexian Li","doi":"10.1016/j.aej.2025.12.069","DOIUrl":"10.1016/j.aej.2025.12.069","url":null,"abstract":"<div><div>Mountain flood disasters in rugged terrains pose significant challenges due to rapid onset, complex spatiotemporal dynamics, and data scarcity, where traditional hydrological models and pairwise graph neural networks struggle to capture multi-scale dependencies and uncertainty in propagation patterns. This study proposes the Adaptive Spatiotemporal Uncertainty-Guided Neural Network (ASTUNN), a hybrid framework that synergistically combines Bidirectional Gated Recurrent Units (BiGRU) for temporal modeling, Spherical Manifold Graph Learning (SMGL) for non-Euclidean spatial analysis, Fractional-Order Dynamic Attention (FODA) for long-memory patterns, Stochastic Variational Inference (SVI) for uncertainty quantification, and Adaptive Feature Synthesis (AFS) for multi-scale fusion. Key innovations include: (1) hyperedge-aware spatiotemporal message passing with fractional-order attention to model higher-order interactions and long-range dependencies in river networks and terrain gradients; and (2) stochastic variational uncertainty estimation to provide calibrated probabilistic forecasts and prevention capability rankings. These contributions overcome limitations of static graphs and deterministic models under rapid environmental changes. Validated on multi-source hydrological datasets from seven high-risk mountainous regions in southwest China, ASTUNN achieves an AUC-ROC of 0.947, MAE of 0.103 for prevention capability rankings, and ECE of 0.029, outperforming state-of-the-art baselines by 15–25 % while reducing false alarms by 18 % and enabling early warnings up to 48 h ahead.</div></div>","PeriodicalId":7484,"journal":{"name":"alexandria engineering journal","volume":"136 ","pages":"Pages 140-156"},"PeriodicalIF":6.8,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145975059","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-01DOI: 10.1016/j.aej.2026.01.005
Yang Xia , Yu Wang , Peng Yu
Online learning platforms’ intelligent feedback mechanisms suffer from static strategies and inadequate content adaptability, failing to meet learners’ dynamic needs. This paper proposes the Dynamic Reinforcement Feedback Network (DRF-Net), integrating Dynamic State Perception, PPO decision-making, and LLaMA 3 generation modules. Experiments on the KDD Cup and OpenEdX datasets show that DRF-Net achieves a learning effect improvement rate of 0.350.05 (34.6% higher than traditional models) and a cumulative reward of 56.84.2 (33.6% higher than single reinforcement learning models). Ablation experiments confirm the necessity of each module — removing the Dynamic State Perception module reduces the learning effect improvement rate by 22.9%. Future work will expand datasets, optimize adaptability to extreme states, and promote the model’s application in real scenarios.
{"title":"A dynamic reinforcement feedback network-based intelligent feedback mechanism in online learning platforms","authors":"Yang Xia , Yu Wang , Peng Yu","doi":"10.1016/j.aej.2026.01.005","DOIUrl":"10.1016/j.aej.2026.01.005","url":null,"abstract":"<div><div>Online learning platforms’ intelligent feedback mechanisms suffer from static strategies and inadequate content adaptability, failing to meet learners’ dynamic needs. This paper proposes the Dynamic Reinforcement Feedback Network (DRF-Net), integrating Dynamic State Perception, PPO decision-making, and LLaMA 3 generation modules. Experiments on the KDD Cup and OpenEdX datasets show that DRF-Net achieves a learning effect improvement rate of 0.35<span><math><mo>±</mo></math></span>0.05 (34.6% higher than traditional models) and a cumulative reward of 56.8<span><math><mo>±</mo></math></span>4.2 (33.6% higher than single reinforcement learning models). Ablation experiments confirm the necessity of each module — removing the Dynamic State Perception module reduces the learning effect improvement rate by 22.9%. Future work will expand datasets, optimize adaptability to extreme states, and promote the model’s application in real scenarios.</div></div>","PeriodicalId":7484,"journal":{"name":"alexandria engineering journal","volume":"136 ","pages":"Pages 30-44"},"PeriodicalIF":6.8,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145975064","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-01DOI: 10.1016/j.aej.2025.12.026
Xiaodan Li , Yue Zhou , Fengchun Gao , Di Cheng , Wushan Li , Kaijian Xia , Hongsheng Yin
Precise prediction of postpartum hemorrhage (PPH) is of great significance for early identification of high-risk pregnant women, optimizing medical resource allocation, and reducing maternal mortality. However, existing PPH prediction methods suffer from limitations such as coarse prediction granularity, and single-stage prediction processes, leading to insufficient prediction accuracy. This has made prediction methods based on hybrid network architectures an important research direction in current PPH studies. This paper proposes a Multi-Granularity Hybrid Network Model (MGHNM) for PPH prediction, which integrates advanced methods such as ensemble learning, convolutional neural networks (CNN), and variational autoencoders (VAE). By leveraging multi-level feature extraction, the model effectively suppresses interference from secondary information, thus significantly enhancing prediction accuracy. The MGHNM model introduces a learnable control switch mechanism to achieve dynamic feature selection, significantly enhancing the model's discriminative ability. By organically combining the CatBoost classifier, CNN feature extractor, VAE representation learning module, and Vision Transformer (ViT), the hybrid network prediction model achieves a significant improvement in prediction accuracy for the three-level classification task of PPH severity (mild, moderate, and severe). The experimental data in this paper is derived from a PPH dataset constructed from the electronic medical record (EMR) system of the Maternal and Child Health Hospital in Jinan, Shandong Province, China. Three experiments were designed: First, the hyperparameters of the prediction model were optimized and analyzed. Second, a multi-model comparative experiment was conducted. Finally, an ablation study was performed. The experimental results demonstrate the significant superiority of the proposed MGHNM model for PPH prediction. It achieves an overall mean accuracy of 89.50 % with a standard deviation of 0.0045 %, substantially outperforming both the baseline and state-of-the-art (SOTA) methods.
{"title":"MGHNM: A multi-granularity based on hybrid network model for postpartum hemorrhage prediction","authors":"Xiaodan Li , Yue Zhou , Fengchun Gao , Di Cheng , Wushan Li , Kaijian Xia , Hongsheng Yin","doi":"10.1016/j.aej.2025.12.026","DOIUrl":"10.1016/j.aej.2025.12.026","url":null,"abstract":"<div><div>Precise prediction of postpartum hemorrhage (PPH) is of great significance for early identification of high-risk pregnant women, optimizing medical resource allocation, and reducing maternal mortality. However, existing PPH prediction methods suffer from limitations such as coarse prediction granularity, and single-stage prediction processes, leading to insufficient prediction accuracy. This has made prediction methods based on hybrid network architectures an important research direction in current PPH studies. This paper proposes a Multi-Granularity Hybrid Network Model (MGHNM) for PPH prediction, which integrates advanced methods such as ensemble learning, convolutional neural networks (CNN), and variational autoencoders (VAE). By leveraging multi-level feature extraction, the model effectively suppresses interference from secondary information, thus significantly enhancing prediction accuracy. The MGHNM model introduces a learnable control switch mechanism to achieve dynamic feature selection, significantly enhancing the model's discriminative ability. By organically combining the CatBoost classifier, CNN feature extractor, VAE representation learning module, and Vision Transformer (ViT), the hybrid network prediction model achieves a significant improvement in prediction accuracy for the three-level classification task of PPH severity (mild, moderate, and severe). The experimental data in this paper is derived from a PPH dataset constructed from the electronic medical record (EMR) system of the Maternal and Child Health Hospital in Jinan, Shandong Province, China. Three experiments were designed: First, the hyperparameters of the prediction model were optimized and analyzed. Second, a multi-model comparative experiment was conducted. Finally, an ablation study was performed. The experimental results demonstrate the significant superiority of the proposed MGHNM model for PPH prediction. It achieves an overall mean accuracy of 89.50 % with a standard deviation of 0.0045 %, substantially outperforming both the baseline and state-of-the-art (SOTA) methods.</div></div>","PeriodicalId":7484,"journal":{"name":"alexandria engineering journal","volume":"135 ","pages":"Pages 371-383"},"PeriodicalIF":6.8,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145923609","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-01DOI: 10.1016/j.aej.2025.12.036
Feryal Abdullah Aladsani , Ali Muhib
In this paper, we focus on finding new oscillation criteria for third-order differential equations. We used a variety of analytical techniques and combined them with new relationships to address some of the problems that have hindered previous studies. As a result, and by using comparability principles, we were able to obtain results that improve and extend some of the previous results published in the literature. We provide some examples to illustrate the effectiveness of the obtained results.
{"title":"New oscillation results for nonlinear delay differential equations of third-order in the canonical case","authors":"Feryal Abdullah Aladsani , Ali Muhib","doi":"10.1016/j.aej.2025.12.036","DOIUrl":"10.1016/j.aej.2025.12.036","url":null,"abstract":"<div><div>In this paper, we focus on finding new oscillation criteria for third-order differential equations. We used a variety of analytical techniques and combined them with new relationships to address some of the problems that have hindered previous studies. As a result, and by using comparability principles, we were able to obtain results that improve and extend some of the previous results published in the literature. We provide some examples to illustrate the effectiveness of the obtained results.</div></div>","PeriodicalId":7484,"journal":{"name":"alexandria engineering journal","volume":"135 ","pages":"Pages 138-143"},"PeriodicalIF":6.8,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145923677","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号