Engineering Applications of Artificial Intelligence最新文献_第5页

K-nearest neighbor-enhanced Residual Learning Framework for image restoration 基于k近邻增强残差学习框架的图像恢复

IF 8 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2026-01-14 DOI: 10.1016/j.engappai.2026.113816

Yaoyun Zeng, Hongxia Wang, Xianchen Zhou

Image restoration is crucial for various applications such as autonomous driving, industrial quality control, and surveillance systems, where clear and reliable images are essential for accurate analysis and decision-making. However, effectively addressing diverse complex degradations, such as blur, missing, and noise, remains a significant challenge during image acquisition. While existing methods have achieved notable progress in handling single degradation, they often lack the flexibility to generalize across different restoration tasks and fail to fully leverage global contextual information and self-similarity patterns within images. To overcome these limitations, we propose a novel Knearest neighbor (KNN)-guided Residual Learning Framework (KRF), specifically designed for restoring images affected by a variety of degradation types using a unified model. The KRF employs an Encoder-Decoder architecture enhanced with strategically placed KNN and residual modules, enabling the network to effectively capture long-range dependencies by leveraging multiscale block-wise relationships. Furthermore, we integrate a spatial pyramid pooling block (SPPB), which improves the network’s robustness by generalizing across varying blur kernels. To optimize feature learning, we design a hybrid loss function that preserves fundamental image features while incorporating Laplacian edge gradients to enhance edge and texture reconstruction. Extensive evaluations demonstrate that our KRF consistently outperforms competitive methods across multiple standard benchmarks.

图像恢复对于自动驾驶、工业质量控制和监控系统等各种应用至关重要，在这些应用中，清晰可靠的图像对于准确的分析和决策至关重要。然而，有效地解决各种复杂的退化，如模糊、缺失和噪声，仍然是图像采集过程中的一个重大挑战。虽然现有方法在处理单个退化方面取得了显著进展，但它们往往缺乏在不同恢复任务之间进行泛化的灵活性，并且不能充分利用图像中的全局上下文信息和自相似模式。为了克服这些限制，我们提出了一种新的最近邻（KNN）引导残差学习框架（KRF），专门用于使用统一模型恢复受各种退化类型影响的图像。KRF采用了一种编码器-解码器架构，通过战略性地放置KNN和残差模块，使网络能够通过利用多尺度块关系有效地捕获远程依赖关系。此外，我们还集成了一个空间金字塔池块（SPPB），通过在不同模糊核之间进行泛化来提高网络的鲁棒性。为了优化特征学习，我们设计了一个混合损失函数，在保留基本图像特征的同时结合拉普拉斯边缘梯度来增强边缘和纹理重建。广泛的评估表明，我们的KRF在多个标准基准测试中始终优于竞争方法。

{"title":"K-nearest neighbor-enhanced Residual Learning Framework for image restoration","authors":"Yaoyun Zeng, Hongxia Wang, Xianchen Zhou","doi":"10.1016/j.engappai.2026.113816","DOIUrl":"10.1016/j.engappai.2026.113816","url":null,"abstract":"<div><div>Image restoration is crucial for various applications such as autonomous driving, industrial quality control, and surveillance systems, where clear and reliable images are essential for accurate analysis and decision-making. However, effectively addressing diverse complex degradations, such as blur, missing, and noise, remains a significant challenge during image acquisition. While existing methods have achieved notable progress in handling single degradation, they often lack the flexibility to generalize across different restoration tasks and fail to fully leverage global contextual information and self-similarity patterns within images. To overcome these limitations, we propose a novel Knearest neighbor (KNN)-guided Residual Learning Framework (KRF), specifically designed for restoring images affected by a variety of degradation types using a unified model. The KRF employs an Encoder-Decoder architecture enhanced with strategically placed KNN and residual modules, enabling the network to effectively capture long-range dependencies by leveraging multiscale block-wise relationships. Furthermore, we integrate a spatial pyramid pooling block (SPPB), which improves the network’s robustness by generalizing across varying blur kernels. To optimize feature learning, we design a hybrid loss function that preserves fundamental image features while incorporating Laplacian edge gradients to enhance edge and texture reconstruction. Extensive evaluations demonstrate that our KRF consistently outperforms competitive methods across multiple standard benchmarks.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"167 ","pages":"Article 113816"},"PeriodicalIF":8.0,"publicationDate":"2026-01-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145979911","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}

引用次数: 0

An adaptive dual-graph spatial–temporal convolutional network with edge-aware fusion for elderly gait recognition using Kinect-based skeleton data 基于kinect骨骼数据的自适应双图时空卷积网络边缘感知融合老年人步态识别

IF 8 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2026-01-13 DOI: 10.1016/j.engappai.2026.113849

Botao Wang , Xiaoyan Wu , Jianning Wu , Qingxin Zeng , Zhaoming Lin

The accurate recognition of elderly gait patterns can significantly contribute to clinical applications such as elderly health monitoring and fall risk prediction. However, how to develop a high-generalization elderly gait classification model has become a challenging problem in elderly gait quantization analysis. Considering the interaction coupling changes across joints in the kinetic chains of elderly gait, we propose an advanced adaptive edge-aware dual-graph convolutional network (AEDGCN) for high-accuracy elderly gait recognition. Our model integrates a gait-graph and a gait-hypergraph to capture high-order joint interaction coupling, which reflects subtle differences in elderly gait changes. By modeling these fine-grained spatial–temporal dependencies, the proposed model achieves strong generalization in accurately identifying elderly gait patterns. Specifically, the proposed technique employs an Edge-Aware Mechanism (EAM) to simultaneously model local spatial dependencies between joints from the gait-graph and cross-joint correlations from the gait-hypergraph. Additionally, the Hierarchical Deep Fully Convolution (HDFC) module is designed to enhance the modeling of temporal dependencies across multiple scales. Our Kinect-based gait dataset, comprising 45 healthy younger participants and 34 healthy elderly participants, with three walking patterns, is used to evaluate the feasibility of our method. In addition, experiments on the public KINECAL dataset further demonstrate the generalization capability of the proposed model. The experimental results confirm that our model outperforms state-of-the-art methods while keeping a low learning complexity. The proposed method effectively enables modeling of elderly gait dynamics, providing informative feature representations for understanding age-related locomotion changes and supporting downstream clinical assessments

准确识别老年人的步态模式对老年人健康监测和跌倒风险预测等临床应用具有重要意义。然而，如何建立一个高泛化的老年人步态分类模型已成为老年人步态量化分析中的一个具有挑战性的问题。针对老年人步态运动链中关节间的交互耦合变化，提出了一种先进的自适应边缘感知双图卷积网络（AEDGCN），用于老年人步态的高精度识别。我们的模型集成了步态图和步态超图来捕捉高阶关节相互作用耦合，这反映了老年人步态变化的细微差异。通过对这些细粒度的时空依赖关系进行建模，该模型在准确识别老年人步态模式方面具有较强的泛化能力。具体而言，该技术采用边缘感知机制（EAM）同时模拟步态图中关节之间的局部空间依赖关系和步态超图中的交叉关节相关性。此外，层次深度全卷积（HDFC）模块旨在增强跨多个尺度的时间依赖性建模。我们的基于运动学的步态数据集包括45名健康的年轻参与者和34名健康的老年人参与者，有三种步行模式，用于评估我们方法的可行性。此外，在KINECAL公共数据集上的实验进一步验证了该模型的泛化能力。实验结果证实，我们的模型在保持较低学习复杂度的同时，优于最先进的方法。所提出的方法有效地实现了老年人步态动力学的建模，为理解与年龄相关的运动变化提供了信息特征表示，并支持下游临床评估

{"title":"An adaptive dual-graph spatial–temporal convolutional network with edge-aware fusion for elderly gait recognition using Kinect-based skeleton data","authors":"Botao Wang , Xiaoyan Wu , Jianning Wu , Qingxin Zeng , Zhaoming Lin","doi":"10.1016/j.engappai.2026.113849","DOIUrl":"10.1016/j.engappai.2026.113849","url":null,"abstract":"<div><div>The accurate recognition of elderly gait patterns can significantly contribute to clinical applications such as elderly health monitoring and fall risk prediction. However, how to develop a high-generalization elderly gait classification model has become a challenging problem in elderly gait quantization analysis. Considering the interaction coupling changes across joints in the kinetic chains of elderly gait, we propose an advanced adaptive edge-aware dual-graph convolutional network (AEDGCN) for high-accuracy elderly gait recognition. Our model integrates a gait-graph and a gait-hypergraph to capture high-order joint interaction coupling, which reflects subtle differences in elderly gait changes. By modeling these fine-grained spatial–temporal dependencies, the proposed model achieves strong generalization in accurately identifying elderly gait patterns. Specifically, the proposed technique employs an Edge-Aware Mechanism (EAM) to simultaneously model local spatial dependencies between joints from the gait-graph and cross-joint correlations from the gait-hypergraph. Additionally, the Hierarchical Deep Fully Convolution (HDFC) module is designed to enhance the modeling of temporal dependencies across multiple scales. Our Kinect-based gait dataset, comprising 45 healthy younger participants and 34 healthy elderly participants, with three walking patterns, is used to evaluate the feasibility of our method. In addition, experiments on the public KINECAL dataset further demonstrate the generalization capability of the proposed model. The experimental results confirm that our model outperforms state-of-the-art methods while keeping a low learning complexity. The proposed method effectively enables modeling of elderly gait dynamics, providing informative feature representations for understanding age-related locomotion changes and supporting downstream clinical assessments</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"167 ","pages":"Article 113849"},"PeriodicalIF":8.0,"publicationDate":"2026-01-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145957653","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}

引用次数: 0

Pixel stable-based novel deep belief network for blood cancer diagnosis 基于像素稳定的新型血癌诊断深度信念网络

IF 8 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2026-01-13 DOI: 10.1016/j.engappai.2025.113676

Fayadh Alenezi

The study explores innovative algorithms for cancer image classification, leveraging advanced metrics to enhance predictive accuracy and minimize diagnostic errors. Accurate classification is critical for timely and effective treatment planning. Traditional deep learning algorithms based on artificial intelligence (AI) often struggle to balance precision, recall, and specificity in image classification tasks, leading to inconsistent results and diagnostic challenges. Addressing these gaps is essential for improving cancer diagnostics. The research introduces machine learning (ML) method based on a deep belief network (DBN) integrating spatial, frequency, and energy metrics. It employs a novel sigmoid activation function and a logistic regression layer, optimized for binary classification, alongside confusion matrix evaluation for precision, recall, specificity, and Area Under the Receiver Operating Characteristic Curve (AUC-ROC) metrics. Achieving 98% accuracy, 99% precision, 98% recall, and 99% specificity, the algorithm demonstrated superior performance. The Formula measure with

β

= 2 (

F_{2}

-score) of 98.2% and AUC-ROC of 98.5% highlight its reliability in cancer classification. The proposed DBN framework offers a significant improvement in cancer image classification, with potential applications in clinical diagnostics and treatment monitoring.

该研究探索了癌症图像分类的创新算法，利用先进的指标来提高预测准确性并最大限度地减少诊断错误。准确的分类对于及时有效的治疗计划至关重要。传统的基于人工智能（AI）的深度学习算法在图像分类任务中往往难以平衡精度、召回率和特异性，从而导致结果不一致和诊断挑战。解决这些差距对于改善癌症诊断至关重要。该研究引入了基于深度信念网络（DBN）的机器学习（ML）方法，该网络集成了空间、频率和能量指标。它采用了一种新的s型激活函数和逻辑回归层，优化了二元分类，以及混淆矩阵评估的精度、召回率、特异性和接收者工作特征曲线下面积（AUC-ROC）指标。该算法的准确率为98%，精密度为99%，召回率为98%，特异性为99%，表现出优异的性能。β = 2 （F2-score）为98.2%，AUC-ROC为98.5%的Formula测量值突出了其在癌症分类中的可靠性。提出的DBN框架在肿瘤图像分类方面有显著的改进，在临床诊断和治疗监测方面具有潜在的应用前景。

{"title":"Pixel stable-based novel deep belief network for blood cancer diagnosis","authors":"Fayadh Alenezi","doi":"10.1016/j.engappai.2025.113676","DOIUrl":"10.1016/j.engappai.2025.113676","url":null,"abstract":"<div><div>The study explores innovative algorithms for cancer image classification, leveraging advanced metrics to enhance predictive accuracy and minimize diagnostic errors. Accurate classification is critical for timely and effective treatment planning. Traditional deep learning algorithms based on artificial intelligence (AI) often struggle to balance precision, recall, and specificity in image classification tasks, leading to inconsistent results and diagnostic challenges. Addressing these gaps is essential for improving cancer diagnostics. The research introduces machine learning (ML) method based on a deep belief network (DBN) integrating spatial, frequency, and energy metrics. It employs a novel sigmoid activation function and a logistic regression layer, optimized for binary classification, alongside confusion matrix evaluation for precision, recall, specificity, and Area Under the Receiver Operating Characteristic Curve (AUC-ROC) metrics. Achieving 98% accuracy, 99% precision, 98% recall, and 99% specificity, the algorithm demonstrated superior performance. The Formula measure with <span><math><mi>β</mi></math></span> = 2 (<span><math><msub><mrow><mi>F</mi></mrow><mrow><mn>2</mn></mrow></msub></math></span>-score) of 98.2% and AUC-ROC of 98.5% highlight its reliability in cancer classification. The proposed DBN framework offers a significant improvement in cancer image classification, with potential applications in clinical diagnostics and treatment monitoring.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"167 ","pages":"Article 113676"},"PeriodicalIF":8.0,"publicationDate":"2026-01-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145957669","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}

引用次数: 0

A deep learning framework for on-street parking demand prediction: Integrating spatio-temporal dynamics and policy impacts 基于深度学习的道路停车需求预测框架：整合时空动态和政策影响

IF 8 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2026-01-13 DOI: 10.1016/j.engappai.2026.113811

Keliang Liu , Jian Chen

Urban on-street short-term parking demand prediction is fundamental for smart parking guidance systems. However, current prediction methods often rely on a single data source and fail to account for the dynamic impacts of environmental factors outside the parking system. This limitation constrains the accuracy of model prediction and does not allow for an assessment of how parking demand is affected by the dynamics of policy changes. To address this issue, this study proposes a comprehensive deep learning forecasting framework. Utilizing data from 123 on-street parking facilities over seven months, totaling more than 6.48 million parking order data. Recognizing the varied spatial and temporal influences that built environment and parking regulations exert on demand patterns, we used the Multi-scale Geographically and Temporally Weighted Regression model (MGTWR) to quantify these relationships. We then incorporated the spatio-temporal coefficients derived from the MGTWR model, alongside additional influential variables, as inputs for a novel deep learning architecture that combines MGTWR, Graph Attention Networks (GAT), and Attention-based Long Short-Term Memory (ALSTM), which we designate as “MGTWR-GAT-ALSTM.” Our model was benchmarked against traditional baseline methods, and the results indicate that MGTWR-GAT-ALSTM yields superior predictive performance, with Mean Absolute Error, Root Mean Squared Error, and Coefficient of Determination metrics of 0.01, 0.04, and 0.92, respectively. Additionally, we performed ablation experiments to confirm that the model design does not introduce redundancy. The proposed prediction model aims to enhance the construction of smart parking systems, providing a dynamic assessment tool for parking policies.

城市街道短期停车需求预测是智能停车引导系统的基础。然而，目前的预测方法往往依赖于单一数据源，无法考虑停车系统外环境因素的动态影响。这种限制限制了模型预测的准确性，并且不允许评估停车需求如何受到政策变化的动态影响。为了解决这个问题，本研究提出了一个全面的深度学习预测框架。利用来自123个街道停车设施的数据，在7个月内，总计超过648万个停车订单数据。认识到建筑环境和停车法规对需求模式的不同时空影响，我们使用多尺度地理和时间加权回归模型（MGTWR）来量化这些关系。然后，我们将来自MGTWR模型的时空系数与其他影响变量结合起来，作为一种新型深度学习架构的输入，该架构结合了MGTWR、图注意网络（GAT）和基于注意的长短期记忆（ALSTM），我们将其命名为“MGTWR-GAT-ALSTM”。我们的模型与传统的基线方法进行了基准测试，结果表明MGTWR-GAT-ALSTM具有优越的预测性能，平均绝对误差、均方根误差和决定系数指标分别为0.01、0.04和0.92。此外，我们进行了烧蚀实验，以确认模型设计不会引入冗余。提出的预测模型旨在加强智能停车系统的建设，为停车政策提供动态评估工具。

{"title":"A deep learning framework for on-street parking demand prediction: Integrating spatio-temporal dynamics and policy impacts","authors":"Keliang Liu , Jian Chen","doi":"10.1016/j.engappai.2026.113811","DOIUrl":"10.1016/j.engappai.2026.113811","url":null,"abstract":"<div><div>Urban on-street short-term parking demand prediction is fundamental for smart parking guidance systems. However, current prediction methods often rely on a single data source and fail to account for the dynamic impacts of environmental factors outside the parking system. This limitation constrains the accuracy of model prediction and does not allow for an assessment of how parking demand is affected by the dynamics of policy changes. To address this issue, this study proposes a comprehensive deep learning forecasting framework. Utilizing data from 123 on-street parking facilities over seven months, totaling more than 6.48 million parking order data. Recognizing the varied spatial and temporal influences that built environment and parking regulations exert on demand patterns, we used the Multi-scale Geographically and Temporally Weighted Regression model (MGTWR) to quantify these relationships. We then incorporated the spatio-temporal coefficients derived from the MGTWR model, alongside additional influential variables, as inputs for a novel deep learning architecture that combines MGTWR, Graph Attention Networks (GAT), and Attention-based Long Short-Term Memory (ALSTM), which we designate as “MGTWR-GAT-ALSTM.” Our model was benchmarked against traditional baseline methods, and the results indicate that MGTWR-GAT-ALSTM yields superior predictive performance, with Mean Absolute Error, Root Mean Squared Error, and Coefficient of Determination metrics of 0.01, 0.04, and 0.92, respectively. Additionally, we performed ablation experiments to confirm that the model design does not introduce redundancy. The proposed prediction model aims to enhance the construction of smart parking systems, providing a dynamic assessment tool for parking policies.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"167 ","pages":"Article 113811"},"PeriodicalIF":8.0,"publicationDate":"2026-01-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145957664","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}

引用次数: 0

Dynamic quantum annealing optimized quantum neural networks for remaining useful lifetime prediction 动态量子退火优化了量子神经网络的剩余有效寿命预测

IF 8 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2026-01-13 DOI: 10.1016/j.engappai.2026.113856

Manoranjan Gandhudi , Gangadharan G.R.

Remaining useful lifetime prediction of complex systems using machine learning remains a significant challenge due to nonlinear degradation dynamics and effective hyperparameter optimization. In this study, we introduce a quantum neural network model optimized through quantum annealing, incorporating a dynamic learning rate strategy to increase prediction accuracy with convergence efficiency. The proposed method adaptively adjusts the learning rate throughout training, facilitating a more effective trade-off between convergence speed and model generalization. We evaluate the model on two benchmark datasets: the Commercial Modular Aero-Propulsion System Simulation (C-MAPSS) dataset which captures turbofan engine degradation patterns and a battery lifetime prediction dataset. Experimental analysis indicates that both dynamic learning rate adaptation and quantum annealing parameter tuning contribute to performance improvement, with annealing alone providing larger gains in certain cases, aligning with our ablation results. The proposed model achieves root mean square error values of 20.29, 26.38, 33.00, and 33.74 across the four subsets of the C-MAPSS dataset and a root mean square error value of 7.01 on the battery lifetime prediction dataset.

由于非线性退化动力学和有效的超参数优化，使用机器学习对复杂系统进行有效的寿命预测仍然是一个重大挑战。在本研究中，我们引入了一个通过量子退火优化的量子神经网络模型，并结合动态学习率策略来提高预测精度和收敛效率。该方法在整个训练过程中自适应调整学习率，在收敛速度和模型泛化之间实现了更有效的权衡。我们在两个基准数据集上对模型进行了评估：商业模块化航空推进系统仿真（C-MAPSS）数据集（捕获涡扇发动机退化模式）和电池寿命预测数据集。实验分析表明，动态学习率自适应和量子退火参数调整都有助于提高性能，在某些情况下，单独退火可以提供更大的增益，这与我们的烧蚀结果一致。该模型在C-MAPSS数据集的四个子集上的均方根误差值分别为20.29、26.38、33.00和33.74，在电池寿命预测数据集上的均方根误差值为7.01。

{"title":"Dynamic quantum annealing optimized quantum neural networks for remaining useful lifetime prediction","authors":"Manoranjan Gandhudi , Gangadharan G.R.","doi":"10.1016/j.engappai.2026.113856","DOIUrl":"10.1016/j.engappai.2026.113856","url":null,"abstract":"<div><div>Remaining useful lifetime prediction of complex systems using machine learning remains a significant challenge due to nonlinear degradation dynamics and effective hyperparameter optimization. In this study, we introduce a quantum neural network model optimized through quantum annealing, incorporating a dynamic learning rate strategy to increase prediction accuracy with convergence efficiency. The proposed method adaptively adjusts the learning rate throughout training, facilitating a more effective trade-off between convergence speed and model generalization. We evaluate the model on two benchmark datasets: the Commercial Modular Aero-Propulsion System Simulation (C-MAPSS) dataset which captures turbofan engine degradation patterns and a battery lifetime prediction dataset. Experimental analysis indicates that both dynamic learning rate adaptation and quantum annealing parameter tuning contribute to performance improvement, with annealing alone providing larger gains in certain cases, aligning with our ablation results. The proposed model achieves root mean square error values of 20.29, 26.38, 33.00, and 33.74 across the four subsets of the C-MAPSS dataset and a root mean square error value of 7.01 on the battery lifetime prediction dataset.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"167 ","pages":"Article 113856"},"PeriodicalIF":8.0,"publicationDate":"2026-01-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145957668","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}

引用次数: 0

Robust planetary gearbox fault diagnosis through time–frequency analysis and transfer learning 基于时频分析和迁移学习的行星齿轮箱鲁棒故障诊断

IF 8 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2026-01-13 DOI: 10.1016/j.engappai.2026.113844

Mahmoud Elhabib Bekaddour Benattia , Houssem Habbouche , Tarak Benkedjouh , Yassine Amirat , Mohamed Benbouzid

Gearboxes are essential mechanical components for power transmission. Among them, planetary gearboxes stand out for their compact design and high reduction ratio, making them particularly advantageous in various sectors such as energy generation, transportation, and robotics. Like all mechanical components under load, they are susceptible to different types of degradation, including wear, cracks, chipping, and even broken teeth. Such defects can negatively impact transmission quality, potentially leading to system shutdowns and endangering operators. This necessitates an autonomous monitoring solution that can respond in real-time. This paper presents a robust monitoring methodology for diagnosing faults in planetary gearboxes. The approach begins with filtering the monitoring signals using Variational Mode Decomposition (VMD). The filtered signal is then transformed into a time–frequency image using Wavelet Transform (WT). These images are subsequently used to train a transfer learning network. To ensure the robustness of the proposed intelligent solution, a data augmentation step is included to address issues of data scarcity and imbalanced datasets. Experimental validation demonstrates the effectiveness of the proposed methodology under different operating conditions.

齿轮箱是动力传动必不可少的机械部件。其中，行星齿轮箱以其紧凑的设计和高减速比脱颖而出，使其在能源生产、交通运输、机器人等各个领域特别具有优势。像所有在载荷下的机械部件一样，它们容易受到不同类型的退化，包括磨损、裂纹、碎裂，甚至断齿。这些缺陷会对传输质量产生负面影响，可能导致系统停机并危及操作人员。这就需要能够实时响应的自主监控解决方案。提出了一种行星齿轮箱故障诊断的鲁棒监测方法。该方法首先使用变分模态分解（VMD）对监测信号进行滤波。然后使用小波变换（WT）将滤波后的信号变换为时频图像。这些图像随后被用于训练迁移学习网络。为了确保所提出的智能解决方案的鲁棒性，包含了一个数据增强步骤来解决数据稀缺和数据集不平衡的问题。实验验证了该方法在不同工况下的有效性。

{"title":"Robust planetary gearbox fault diagnosis through time–frequency analysis and transfer learning","authors":"Mahmoud Elhabib Bekaddour Benattia , Houssem Habbouche , Tarak Benkedjouh , Yassine Amirat , Mohamed Benbouzid","doi":"10.1016/j.engappai.2026.113844","DOIUrl":"10.1016/j.engappai.2026.113844","url":null,"abstract":"<div><div>Gearboxes are essential mechanical components for power transmission. Among them, planetary gearboxes stand out for their compact design and high reduction ratio, making them particularly advantageous in various sectors such as energy generation, transportation, and robotics. Like all mechanical components under load, they are susceptible to different types of degradation, including wear, cracks, chipping, and even broken teeth. Such defects can negatively impact transmission quality, potentially leading to system shutdowns and endangering operators. This necessitates an autonomous monitoring solution that can respond in real-time. This paper presents a robust monitoring methodology for diagnosing faults in planetary gearboxes. The approach begins with filtering the monitoring signals using Variational Mode Decomposition (VMD). The filtered signal is then transformed into a time–frequency image using Wavelet Transform (WT). These images are subsequently used to train a transfer learning network. To ensure the robustness of the proposed intelligent solution, a data augmentation step is included to address issues of data scarcity and imbalanced datasets. Experimental validation demonstrates the effectiveness of the proposed methodology under different operating conditions.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"167 ","pages":"Article 113844"},"PeriodicalIF":8.0,"publicationDate":"2026-01-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145957663","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}

引用次数: 0

An electroencephalogram signal analysis method based on dual self-supervised graph diffusion recurrent network 基于对偶自监督图扩散递归网络的脑电图信号分析方法

IF 8 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2026-01-13 DOI: 10.1016/j.engappai.2026.113820

Sunan Ge , Shuang Wang , Rui Zhang , Xueqing Zhao , Xinshi , Meng Wang , Tao Wu

Diagnosis of neurological diseases and emotion recognition analyzing based on electroencephalogram (EEG) signals have been widely applied in numerous fields by revealing the complex operational mechanisms of the human brain. However, existing EEG signal analysis methods are hindered by label noise and the scale of labeled data samples, making it difficult to effectively learn the distribution characteristics of the data and identify the heterogeneity of EEG signals. Therefore, this paper proposes a dual self-supervised graph diffusion recurrent network (DSGDRN) method for representation learning of unlabeled EEG signals, reducing biases and noise effects caused by manual annotation and improving the ability to recognize individual differences. First, to capture the natural geometric features of EEG signals and the dynamic connection information within the brain, distance graph structures and correlation graph structures are respectively used for feature expression. A dual self-supervised algorithm is employed to represent hidden states as a learnable function, enhancing the expressive power of the graph recurrent diffusion network and its ability to recognize contextual information. Finally, during the testing process, a dual learning strategy with continuous adaptive adjustment of hidden state parameters is adopted to improve the application capability of EEG signals in real-world scenarios. Experimental results demonstrate that compared with existing methods, the proposed method exhibits superior performance in neurological disease diagnosis and emotion detection, indicating its effective representation learning capabilities in fields such as neurological disease analysis and emotion recognition.

基于脑电图（EEG）信号的神经系统疾病诊断和情绪识别分析通过揭示人脑复杂的运行机制已广泛应用于众多领域。然而，现有的脑电信号分析方法受到标注噪声和标注数据样本规模的阻碍，难以有效地了解数据的分布特征，识别脑电信号的异质性。因此，本文提出了一种双自监督图扩散递归网络（DSGDRN）方法，用于未标记脑电信号的表示学习，减少人工标注带来的偏差和噪声影响，提高个体差异的识别能力。首先，利用距离图结构和关联图结构分别进行特征表达，捕捉脑电信号的自然几何特征和脑内的动态连接信息；采用对偶自监督算法将隐藏状态表示为可学习函数，增强了图递归扩散网络的表达能力和对上下文信息的识别能力。最后，在测试过程中，采用连续自适应调整隐藏状态参数的双重学习策略，提高脑电信号在实际场景中的应用能力。实验结果表明，与现有方法相比，该方法在神经系统疾病诊断和情绪检测方面表现出优越的性能，表明其在神经系统疾病分析和情绪识别等领域具有有效的表征学习能力。

{"title":"An electroencephalogram signal analysis method based on dual self-supervised graph diffusion recurrent network","authors":"Sunan Ge , Shuang Wang , Rui Zhang , Xueqing Zhao , Xinshi , Meng Wang , Tao Wu","doi":"10.1016/j.engappai.2026.113820","DOIUrl":"10.1016/j.engappai.2026.113820","url":null,"abstract":"<div><div>Diagnosis of neurological diseases and emotion recognition analyzing based on electroencephalogram (EEG) signals have been widely applied in numerous fields by revealing the complex operational mechanisms of the human brain. However, existing EEG signal analysis methods are hindered by label noise and the scale of labeled data samples, making it difficult to effectively learn the distribution characteristics of the data and identify the heterogeneity of EEG signals. Therefore, this paper proposes a dual self-supervised graph diffusion recurrent network (DSGDRN) method for representation learning of unlabeled EEG signals, reducing biases and noise effects caused by manual annotation and improving the ability to recognize individual differences. First, to capture the natural geometric features of EEG signals and the dynamic connection information within the brain, distance graph structures and correlation graph structures are respectively used for feature expression. A dual self-supervised algorithm is employed to represent hidden states as a learnable function, enhancing the expressive power of the graph recurrent diffusion network and its ability to recognize contextual information. Finally, during the testing process, a dual learning strategy with continuous adaptive adjustment of hidden state parameters is adopted to improve the application capability of EEG signals in real-world scenarios. Experimental results demonstrate that compared with existing methods, the proposed method exhibits superior performance in neurological disease diagnosis and emotion detection, indicating its effective representation learning capabilities in fields such as neurological disease analysis and emotion recognition.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"167 ","pages":"Article 113820"},"PeriodicalIF":8.0,"publicationDate":"2026-01-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145957670","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}

引用次数: 0

A novel two-stage intelligent Kalman filter for maneuvering target tracking 一种用于机动目标跟踪的两级智能卡尔曼滤波器

IF 8 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2026-01-13 DOI: 10.1016/j.engappai.2026.113843

Benqi Zhao , Gaoliang Peng , Wei Zhang , Jinghan Wang , Shiji Zhang , Feng Cheng

As the threat posed by unmanned aerial vehicles (UAVs) continues to grow, the integration of optoelectronic tracking systems with laser emitters has emerged as the most advanced solution. However, in the presence of frequent UAV maneuvers and environmental disturbances, current tracking approaches struggle to ensure both stability and accuracy, which compromises the ability to maintain sustained laser focus on the target. To address this challenge, this paper proposes a novel Two-Stage Intelligent Kalman Filter (TSIKF). By modeling target motion dynamics and environmental disturbance features through dedicated neural networks, we achieve a structured decomposition of uncertainties during tracking. Furthermore, we propose dynamic parameter embedding methods that map feature parameters in real time to the state space model, enabling adaptive, interpretable, and high-precision tracking in complex scenarios. The proposed method is rigorously validated through multiple experiments, including a simulation example, two publicly available UAV datasets, and a real-world tracking system. In all tests, TSIKF consistently outperforms existing state-of-the-art methods in terms of accuracy, stability, and generalizability, while fully meeting the real-time processing requirement. Experimental results demonstrate that TSIKF significantly enhances the system’s ability to maintain focus. This work presents an effective approach to high-dynamic target tracking, and provides theoretical insights and technical support for the development of next-generation intelligent optoelectronic interception systems

随着无人机带来的威胁不断增长，光电跟踪系统与激光发射器的集成已经成为最先进的解决方案。然而，在频繁的无人机机动和环境干扰的存在下，目前的跟踪方法努力确保稳定性和准确性，这损害了保持持续激光聚焦在目标上的能力。为了解决这一挑战，本文提出了一种新的两阶段智能卡尔曼滤波器（TSIKF）。通过专用神经网络对目标运动动力学和环境干扰特征进行建模，实现了跟踪过程中不确定性的结构化分解。此外，我们提出了动态参数嵌入方法，将特征参数实时映射到状态空间模型，实现复杂场景下的自适应、可解释和高精度跟踪。该方法通过多个实验进行了严格验证，包括一个仿真示例、两个公开可用的无人机数据集和一个真实世界的跟踪系统。在所有测试中，TSIKF在准确性，稳定性和通用性方面始终优于现有的最先进的方法，同时完全满足实时处理要求。实验结果表明，TSIKF显著提高了系统保持焦点的能力。该研究为高动态目标跟踪提供了一种有效的方法，为下一代智能光电拦截系统的发展提供了理论见解和技术支持

{"title":"A novel two-stage intelligent Kalman filter for maneuvering target tracking","authors":"Benqi Zhao , Gaoliang Peng , Wei Zhang , Jinghan Wang , Shiji Zhang , Feng Cheng","doi":"10.1016/j.engappai.2026.113843","DOIUrl":"10.1016/j.engappai.2026.113843","url":null,"abstract":"<div><div>As the threat posed by unmanned aerial vehicles (UAVs) continues to grow, the integration of optoelectronic tracking systems with laser emitters has emerged as the most advanced solution. However, in the presence of frequent UAV maneuvers and environmental disturbances, current tracking approaches struggle to ensure both stability and accuracy, which compromises the ability to maintain sustained laser focus on the target. To address this challenge, this paper proposes a novel Two-Stage Intelligent Kalman Filter (TSIKF). By modeling target motion dynamics and environmental disturbance features through dedicated neural networks, we achieve a structured decomposition of uncertainties during tracking. Furthermore, we propose dynamic parameter embedding methods that map feature parameters in real time to the state space model, enabling adaptive, interpretable, and high-precision tracking in complex scenarios. The proposed method is rigorously validated through multiple experiments, including a simulation example, two publicly available UAV datasets, and a real-world tracking system. In all tests, TSIKF consistently outperforms existing state-of-the-art methods in terms of accuracy, stability, and generalizability, while fully meeting the real-time processing requirement. Experimental results demonstrate that TSIKF significantly enhances the system’s ability to maintain focus. This work presents an effective approach to high-dynamic target tracking, and provides theoretical insights and technical support for the development of next-generation intelligent optoelectronic interception systems</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"167 ","pages":"Article 113843"},"PeriodicalIF":8.0,"publicationDate":"2026-01-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145957671","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}

引用次数: 0

Robust synchronization of chaotic systems using noise-resistant gradient neural dynamics: Design and application 基于抗噪声梯度神经动力学的混沌系统鲁棒同步：设计与应用

IF 8 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2026-01-13 DOI: 10.1016/j.engappai.2026.113854

Guancheng Wang , Liu Yang , Fenghao Zhuang , Lingbo Han , Zhihao Hao , Xiuchun Xiao , Cong Lin

The synchronization of chaotic systems has found extensive applications in various fields, including secure communication, financial modeling, and image encryption. However, in practical scenarios, chaotic system dynamics are often significantly affected by external noise, which degrades trajectory stability and adversely impacts synchronization performance. Addressing noise-induced degradation has therefore become a critical research focus in chaotic system studies. As a key area of artificial intelligence, neural dynamics (ND) plays a significant role in modeling and optimizing complex systems. In this paper, a robust controller is designed to address this issue within a general master–slave chaotic system framework by employing Noise-Resistant Gradient Neural Dynamics (NRGND), which effectively mitigates the effects of noise and enhances synchronization efficacy. Notably, the controller requires only that the master and slave systems share the same dimensionality and that their system states and corresponding time derivatives are observable. In addition, theoretical analyses of the controller under various noises are conducted, demonstrating its exceptional convergence and robustness. Next, experiments conducted in representative chaotic systems under various noisy scenarios illustrate the superior performance of the NRGND-based controller in achieving synchronization. Finally, the performance of the NRGND controller was evaluated in two applications. The proposed method achieves image encryption and decryption by driving the Lorenz and Lu chaotic systems into synchronization. Furthermore, the NRGND-based robotic motion control scheme demonstrated its robustness and stability under noisy conditions, highlighting its potential for real-world engineering applications.

混沌系统的同步在安全通信、金融建模和图像加密等领域有着广泛的应用。然而，在实际应用中，混沌系统动力学通常会受到外界噪声的显著影响，从而降低轨迹稳定性并对同步性能产生不利影响。因此，解决噪声引起的退化问题已成为混沌系统研究中的一个重要研究热点。神经动力学作为人工智能的一个重要研究领域，在复杂系统建模和优化方面发挥着重要作用。本文采用抗噪声梯度神经动力学（NRGND）设计了一种鲁棒控制器，在一般主从混沌系统框架下解决了这一问题，有效地减轻了噪声的影响，提高了同步效率。值得注意的是，控制器只要求主系统和从系统共享相同的维度，并且它们的系统状态和相应的时间导数是可观察的。此外，对该控制器在各种噪声下的性能进行了理论分析，证明了该控制器具有良好的收敛性和鲁棒性。接下来，在具有代表性的混沌系统中进行了各种噪声场景下的实验，验证了基于nrgnd的控制器在实现同步方面的优越性能。最后，在两个应用中对NRGND控制器的性能进行了评价。该方法通过驱动洛伦兹混沌系统和鲁混沌系统同步实现图像加解密。此外，基于nrgnd的机器人运动控制方案在噪声条件下具有鲁棒性和稳定性，突出了其在实际工程应用中的潜力。

{"title":"Robust synchronization of chaotic systems using noise-resistant gradient neural dynamics: Design and application","authors":"Guancheng Wang , Liu Yang , Fenghao Zhuang , Lingbo Han , Zhihao Hao , Xiuchun Xiao , Cong Lin","doi":"10.1016/j.engappai.2026.113854","DOIUrl":"10.1016/j.engappai.2026.113854","url":null,"abstract":"<div><div>The synchronization of chaotic systems has found extensive applications in various fields, including secure communication, financial modeling, and image encryption. However, in practical scenarios, chaotic system dynamics are often significantly affected by external noise, which degrades trajectory stability and adversely impacts synchronization performance. Addressing noise-induced degradation has therefore become a critical research focus in chaotic system studies. As a key area of artificial intelligence, neural dynamics (ND) plays a significant role in modeling and optimizing complex systems. In this paper, a robust controller is designed to address this issue within a general master–slave chaotic system framework by employing Noise-Resistant Gradient Neural Dynamics (NRGND), which effectively mitigates the effects of noise and enhances synchronization efficacy. Notably, the controller requires only that the master and slave systems share the same dimensionality and that their system states and corresponding time derivatives are observable. In addition, theoretical analyses of the controller under various noises are conducted, demonstrating its exceptional convergence and robustness. Next, experiments conducted in representative chaotic systems under various noisy scenarios illustrate the superior performance of the NRGND-based controller in achieving synchronization. Finally, the performance of the NRGND controller was evaluated in two applications. The proposed method achieves image encryption and decryption by driving the Lorenz and Lu chaotic systems into synchronization. Furthermore, the NRGND-based robotic motion control scheme demonstrated its robustness and stability under noisy conditions, highlighting its potential for real-world engineering applications.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"167 ","pages":"Article 113854"},"PeriodicalIF":8.0,"publicationDate":"2026-01-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145957695","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}

引用次数: 0

Enhanced training data acquisition system for artificial intelligence-enabled camera in smartphones 增强了智能手机中启用人工智能相机的训练数据采集系统

IF 8 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2026-01-13 DOI: 10.1016/j.engappai.2026.113736

Kyeongjun Kim , Youngjo Kim , Hyunhee Park , Dongweon Yoon

With the rapid advancement of deep learning (DL) within the field of artificial intelligence (AI), computer vision technologies have been increasingly integrated into smartphone cameras. Developing DL-based solutions for AI-enabled smartphone cameras, however, demands large volumes of high-quality training data. To address this challenge, this paper introduces a Dual-Camera Real-Image (DCRI) data acquisition system and demonstrates that pretrained networks can be further enhanced through fine-tuning on the proposed dataset. Specifically, the DCRI system comprises a beam splitter, a smartphone camera, and a high-performance digital single-lens reflex (DSLR) camera. We also propose a post-processing pipeline that aligns and color-corrects the paired images, effectively resolving the alignment difficulties commonly observed in prior methods. Extensive experiments confirm that models trained on the DCRI dataset for deep learning-based image signal processing (DL-ISP) achieve substantial improvements in image detail and noise reduction compared with existing approaches. The proposed dataset is publicly available for download.¹

随着人工智能（AI）领域深度学习（DL）的快速发展，计算机视觉技术已经越来越多地集成到智能手机相机中。然而，为支持人工智能的智能手机摄像头开发基于dl的解决方案需要大量高质量的训练数据。为了解决这一挑战，本文介绍了一种双摄像头实时图像（DCRI）数据采集系统，并证明通过对所提出的数据集进行微调，可以进一步增强预训练网络。具体来说，DCRI系统包括一个分束器、一个智能手机相机和一个高性能数字单反（DSLR）相机。我们还提出了一个后处理管道，对配对图像进行对齐和颜色校正，有效地解决了先前方法中常见的对齐困难。大量实验证实，与现有方法相比，在基于深度学习的图像信号处理（DL-ISP）的DCRI数据集上训练的模型在图像细节和降噪方面取得了实质性的改进。建议的数据集可公开下载

{"title":"Enhanced training data acquisition system for artificial intelligence-enabled camera in smartphones","authors":"Kyeongjun Kim , Youngjo Kim , Hyunhee Park , Dongweon Yoon","doi":"10.1016/j.engappai.2026.113736","DOIUrl":"10.1016/j.engappai.2026.113736","url":null,"abstract":"<div><div>With the rapid advancement of deep learning (DL) within the field of artificial intelligence (AI), computer vision technologies have been increasingly integrated into smartphone cameras. Developing DL-based solutions for AI-enabled smartphone cameras, however, demands large volumes of high-quality training data. To address this challenge, this paper introduces a <em>Dual-Camera Real-Image (DCRI)</em> data acquisition system and demonstrates that pretrained networks can be further enhanced through fine-tuning on the proposed dataset. Specifically, the DCRI system comprises a beam splitter, a smartphone camera, and a high-performance digital single-lens reflex (DSLR) camera. We also propose a post-processing pipeline that aligns and color-corrects the paired images, effectively resolving the alignment difficulties commonly observed in prior methods. Extensive experiments confirm that models trained on the DCRI dataset for deep learning-based image signal processing (DL-ISP) achieve substantial improvements in image detail and noise reduction compared with existing approaches. The proposed dataset is publicly available for download.<span><span><sup>1</sup></span></span></div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"167 ","pages":"Article 113736"},"PeriodicalIF":8.0,"publicationDate":"2026-01-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145957665","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}

引用次数: 0