检索结果-内蒙古大学图书馆

2025 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2025

作者： Chen, Yuxin Lin, Fangru Huo, Jingyi Yan, Hui School of Computer Science and Engineering Nanjing University of Science and Technology Nanjing China

ISBN: (纸本)9798350368741

Multivariate Time-Series (MTS) forecasting is challenging due to the complex spatial-temporal dependencies inherent in MTS data. Recent studies typically adopt spatial-temporal graph models to leverage this information. However, most of these approaches assume homophily in graphs and perform only immediate message passing, which limits their ability to fully capture spatial-temporal dependencies. In this paper, we propose a novel Dual-Path Spatial-Temporal Graph Neural Network (DPSTGNN). We first extract and cluster representative short-term series patterns from the MTS data. Then, by comparing historical series segments with the extracted representative patterns, we generate pattern attribution embeddings for nodes, enabling the construction of Homophily and Heterophily Spatial-Temporal Crossed Graphs (STCGs), which depict diverse node-wise relationships across different time segments. Finally, we introduce dilated spatial-temporal crossed graph convolutions on both types of STCGs to capture a wide range of spatial-temporal dependencies for enhanced MTS forecasting. Experiments on six real-world datasets demonstrate that our method outperforms state-of-the-art models. © 2025 IEEE.

关键词： Graph Neural Network Heterophily Multivariate Time-Series Forecasting

来源：评论

学校读者我要写书评

暂无评论

QRALadder: QoE and Resource Consumption-Aware Encoding Ladder Optimization for Live Video Streaming 31st

QRALadder: QoE and Resource Consumption-Aware Encoding Ladd...

引用

31st International Conference on Multimedia Modeling, MMM 2025

作者： Zhu, Yingqian Gao, Guanyu School of Computer Science and Engineering Nanjing University of Science and Technology Nanjing China

ISBN: (纸本)9789819620630

Live video streaming demands high user Quality of Experience (QoE) and requires significant computing power and bandwidth for video encoding and transmission. The standard adaptive live streaming approach encodes the source video content at multiple predefined bitrates, allowing users to select the most appropriate bitrate for optimal rate adaptation. The current methods for encoding ladder design primarily focus on optimizing user QoE. However, the relationship between the encoding ladder and user QoE is complex due to the dynamic nature of live streaming, where the number of users fluctuates, and each user experiences varying and evolving bandwidth conditions. On the other hand, resource consumption in live streaming significantly impacts operational costs, as it requires substantial resources for both video encoding and delivery. To address this challenge, we propose QRALadder, a QoE and resource consumption-aware approach to encoding ladder design. Our method simultaneously optimizes user QoE while considering the impact of the encoding ladder on resource consumption in live streaming, and it can dynamically select the most suitable encoding ladder based on the current system state. We use the Dueling Double Deep Q Networks (Dueling DDQN) algorithm to learn the optimal encoding ladder, maximizing overall performance by analyzing user QoE feedback and resource consumption about system states. We conduct comparative experiments using various types of video content and real-world bandwidth traces to validate our approach. The results show that QRALadder can achieve higher overall performance compared to baseline methods. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Bandwidth

来源：评论

学校读者我要写书评

暂无评论

High-Efficiency Asymmetrically Designed Three-Phase Virtual 48-pulse Power Supply for Electrolytic Hydrogen

引用

IEEE Transactions on Power Electronics 2025年第8期40卷 10335-10342页

作者： Tian, Hanlei Cui, Haoran Han, Wei Sathik M, Jagabar Liang, Guozhuang Chen, Maolin Mekhilef, Saad Sustainable Energy and Environment Thrust The Hong Kong University of Science and Technology (GZ) Guangzhou China Electrical Engineering School of Hebei University of Science and Technology Shijiazhuang China Department of Electronic and Computer Engineering The Hong Kong University of Science and Technology Hong Kong Faculty of Engineering and Technology SRM Institute of Science and Technology Kattankulathur India College of Electrical Engineering Sichuan University Chengdu China School of Science Computing and Engineering Technologies Swinburne University of Technology Melbourne VIC Australia

To address the low-voltage, high-current requirements in hydrogen production applications, a virtual 48-pulse three-phase rectifier is proposed, and achieves the equivalent performance of four parallel 12-pulse rectifiers (12-PR). The proposed approach builds upon the conventional 12-PR design and incorporates novel Asymmetric Current Injection Units (ACIU). The ACIU is installed on the DC side of the AC-DC converter, generating a specific current at the DC link to modulate and enhance the operating modes of the 12-PR. Consequently, the 12-pulse is extended to 48, resulting in an almost sinusoidal input current with a Total Harmonic Distortion (THD) of 3.2% and an output current with a ripple rate (RR) of 2.3%, thus significantly improving power quality. Furthermore, the asymmetric design of the current injection units eliminates the need for a phase-shifting transformer. The efficacy of the proposed rectifier for electrolytic hydrogen production is validated using a 2.2 kW experimental prototype. © 1986-2012 IEEE.

关键词： Voltage Hydrogen Rectifiers Production Windings Circuits Inductors Transformers Harmonic analysis Electrochemical processes

来源：评论

学校读者我要写书评

暂无评论

Congestion Based Adaptive Association Mechanism for IEEE 802.11ah Based Large IoT Network 5th

Congestion Based Adaptive Association Mechanism for IEEE 80...

引用

5th EAI International Conference on Cognitive Computing and Cyber Physical Systems, IC4S 2024

作者： R, Nandhini R, Radha School of Computer Science and Engineering Vellore Institute of Technology–Chennai Chennai India

ISBN: (纸本)9783031770807

The IEEE 802.11 ah standard is the prominent protocol for medium range communication in the domain of Internet of Things (IoT), facilitating interaction between wireless sensor nodes and access points. Supporting up to approximately 8192 nodes connected to a single access point through single hop strategies, IEEE 802.11 ah employs CSMA/CA for the authentication and association process of sensor nodes with access point. However, the standard encounters challenges such as significant collisions and prolonged network setup times when attempting to associate an excessive number of wireless sensor nodes with a single access point, primarily due to authentication thresholds set at intervals. In this research, we propose a dynamic adjustment of the authentication threshold based on network congestion status. Learning parameters including the number of requests served successfully by the access point in the past, present and a congestion threshold are leveraged to adapt to current network congestion levels. Through evaluation using the LILD method in the ns-3 simulator, our approach demonstrates considerable enhancements in terms of collision reduction and network setup efficiency. This research endeavours to promote responsible network management practices for improving the performance and reliability of IEEE 802.11 ah based IoT deployments. © ICST Institute for computer Sciences, Social Informatics and Telecommunications engineering 2025.

关键词： Internet of things

来源：评论

学校读者我要写书评

暂无评论

HyPRETo: Hybrid Pre-trained Ontology Approach for Contextual Relation Classification on Mosquito Vector Biocontrol Agents 8th

HyPRETo: Hybrid Pre-trained Ontology Approach for Contextual...

引用

8th IFIP TC 12 International Conference on computer, Communication and Signal Processing, ICCCSP 2024

作者： Jeyakodi, G. Bala, P. Shanthi Department of Computer Science School of Engineering and Technology Pondicherry University Puducherry India

ISBN: (纸本)9783031736162

Pre-trained Language Model facilitates contextual relation classification by capturing contextual information, addressing word ambiguity, encoding global sentence context, enabling transfer learning, handling out-of-vocabulary words, and improving performance with limited labelled data. Existing pre-training approaches suffer in size, bias, interpretability, generalization, and the lack of domain specificity. To address this, HyPRETo, the hybrid model that combines the strength of token replacement and dynamic masking is proposed to achieve upgraded performance to increase classification accuracy. The Mosquito Vector Biocontrol Agents data is used for implementing the model for a contextual relation classification task. HyPRETo uses ontology to provide structured knowledge. HyPRETo is pre-trained by ELECTRA and fine-tuned by RoBERTa models. Feedforward and softmax activation function is used for classification. The Natural Language Processing technique and SQL database are used to develop an automated question-answering system. The HyPRETo was evaluated with state-of-art models and achieved 98.42% accuracy. As a contribution, the manually annotated input dataset on the mosquito vector control agent is prepared for the classification task. Subsequently, the enhanced model is developed. The interface for an automated question-answering system for mosquito vector biocontrol agents is developed to assist public health applications such as mosquito vector control, disease control, ecosystem management, environmental conservation, and so on. © IFIP International Federation for Information Processing 2025.

关键词： Vectors

来源：评论

学校读者我要写书评

暂无评论

Data-Driven Characterization and Forecasting of Metal-Oxide Varistor Degradation in DC Circuit Breakers 14

Data-Driven Characterization and Forecasting of Metal-Oxide ...

引用

14th Annual IEEE Applied Power Electronics Conference and Exposition, APEC 2025

作者： Zhang, Zhi Jin Justin Liu, Yang Graber, Lukas Saeedifard, Maryam Georgia Institute of Technology School of Electrical and Computer Engineering Atlanta United States

ISBN: (纸本)9798331516116

Energy absorption devices play a critical role in semiconductor-based dc circuit breakers (DCCBs), where they generate a counter-voltage to reverse fault current direction and dissipate fault energy during the interruption process. The metal-oxide varistor (MOV) is the mainstream candidate for this role. However, since MOVs are primarily designed for electrical surge protection, their degradation patterns under dc fault interruption conditions, characterized by exposure to unique DCCB voltage and current waveforms, have not been thoroughly studied. Understanding and predicting MOV degradation (an indicator of MOV lifetime) is crucial for ensuring the reliability and longevity of DCCBs. Therefore, there is a need to (i) systematically test and characterize the degradation of MOVs during dc fault interruptions and (ii) develop forecasting methods to estimate their long-term degradation. In this work, MOV degradation data is collected using an experimental DCCB testbed designed for accelerated degradation testing. Based on this data, statistical analysis and machine learning techniques are then employed to analyze the data, providing insight into DCCB MOV degradation trends and enabling the forecasting of their long-term performance. This work lays the foundation for enhancing the operational reliability and life-cycle management of DCCBs. © 2025 IEEE.

关键词： Varistors

来源：评论

学校读者我要写书评

暂无评论

Research on vehicle detection algorithm based on drone image

Research on vehicle detection algorithm based on drone image

引用

2024 International Conference on Physics, Photonics, and Optical engineering, ICPPOE 2024

作者： Yu, Xiang Yang, Wu School of Computer Science and Engineering Chongqing University of Technology Chongqing404046 China

ISBN: (数字)9781510689121

ISBN: (纸本)9781510689114

To improve the accuracy of vehicle detection, this paper proposes an intelligent model to meet the real-time needs of vehicle detection based on the current common drone shooting methods. To reduce the computational parameters in the feature extraction layer, this paper uses depth separable convolution as the feature extractor. Compared with the conventional convolution kernel that processes the spatial and channel features of the feature map at the same time, the depth separable convolution kernels use depth convolution and point-by-point convolution to extract and learn the spatial and channel features of the image respectively. In addition, aiming at the problems of existing deep learning object detection models such as high hardware requirements and unstable cloud server detection performance, this paper proposes a lightweight end-to-end vehicle detection model that can be deployed on drone embedded hardware terminals and perform real-time inference. Finally, the reliability of the proposed model is verified by experimental research, thus providing a reliable technical reference for subsequent vehicle detection. © 2025 SPIE.

关键词： Drones

来源：评论

学校读者我要写书评

暂无评论

Dynamic Modeling and Analysis for Electricity-Gas Systems With Electric-Driven Compressors

引用

IEEE Transactions on Smart Grid 2025年第3期16卷 2144-2155页

作者： Huang, Yujia Sun, Qiuye Chen, Zhe Wenzhong Gao, David Bach Pedersen, Torben Guldstrand Larsen, Kim Li, Yushuai Shenyang University of Technology School of Electrical Engineering Shenyang110870 China Northeastern University School of Information Science and Engineering Shenyang110819 China Aalborg University Department of Energy Technology Aalborg9220 Denmark University of Denver Department of Electrical and Computer Engineering DenverCO80208 United States Aalborg University Department of Computer Science Aalborg9220 Denmark

The dynamic gas flow model with static electric driven compressor (EDC) model has been widely studied in coordinated analysis of integrated electricity-gas system (IEGS). However, as a crucial coupling unit, the boost characteristics of the EDC change dynamically with gas state, which may lead to unstable operation mode and then affect safe operating of IEGS. Meanwhile, its nonlinear model greatly increases the difficulty of analysis. This paper proposes a reduced-order transfer function model that considers EDCs for an accurate and efficient dynamic analysis of the IEGS. Firstly, a general two-port model in the Laplace-domain is derived to jointly analyze the pipeline and EDCs dynamic characteristics. The model based on transfer function, explicitly gives the relation between the pressures and flows state at ports. On this basis, to concisely analyze state variations of crucial nodes, the multiple cascaded pipelines and compressors are aggregated and to form an equivalent model. Meanwhile, to reduce the computational burden of the high-order Laplacian "s" in the aggregated process, a reduced-order method is developed based on Jordan continued-fraction expansion by preserving the main gas dynamic feature. Then the analytical expression of state fluctuation can be conveniently derived to analyze the underlying impact and interaction between power system and gas system. Finally, case studies are conducted to prove the effectiveness of proposed model and reduction method. © 2010-2012 IEEE.

关键词： Compressibility of gases

来源：评论

学校读者我要写书评

暂无评论

A novel composite machine learning model for the prediction of compressive strength of blended concrete

引用

Journal of Building Pathology and Rehabilitation 2025年第1期10卷 1-18页

作者： Prasad, E.V. Rama Krishna, S. Singha, S. Department of Civil Engineering School of Technology GITAM (Deemed to be University) Telangana Hyderabad India Department of Computer Science and Engineering School of Technology GITAM (Deemed to be University) Telangana Hyderabad India

Blended cement, comprising clinker and supplementary cementitious materials (SCMs), such as fly ash, slag, and silica fume, forms blended concrete when combined with aggregates. This study introduced a novel ensemble approach, the LightGBM and ANN Fusion (LAF) model, to check the prediction of compressive strength in blended concrete using eight key input features, including cement content, slag, fly ash, water, and curing age, derived from 1030 samples. The LAF model synergistically combines predictions from LightGBM and ANN with the fusion weight (alpha) dynamically optimized using Support Vector Machine (SVM) analysis. This integration leverages the strengths of both models while mitigating their individual limitations. The LAF model exhibited superior performance, with an R² of 0.9514 during training and 0.8863 during testing, significantly surpassing standalone models such as LightGBM (R² = 0.9385) and ANN (R² = 0.9136). It achieved the lowest error rates, including an MSE of 15.7384 and RMSE of 3.9672 during training, underscoring its robust generalization capabilities. The novel utilization of SVM-tuned alpha for dynamic model adaptation reduces overfitting and enhances predictive accuracy compared to traditional ensemble and single models. Comparative analysis with existing ensemble techniques, including stacking and gradient-boosted ANN, further validated the superior accuracy and adaptability of the LAF model. This approach represents a significant advancement in predictive modeling, offering a scientifically sound and versatile framework for concrete strength prediction. Future research will refine the SVM tuning, expand the model’s application to other material properties, and explore real-time deployment, positioning the LAF model as a pioneering tool for smart construction materials. © The Author(s), under exclusive licence to Springer Nature Switzerland AG 2024.

关键词： ANN Compressive strength LAF LGBM Performance indicators Sensitivity analysis SVM

来源：评论

学校读者我要写书评

暂无评论

Collision Avoidance System using YOLO-based Object Detection and Distance Estimation 3

Collision Avoidance System using YOLO-based Object Detection...

引用

3rd International Conference on Intelligent Data Communication Technologies and Internet of Things, IDCIoT 2025

作者： Raj, Shiv Akash Dhinakar Singh, Ashita Shekhawat, Tarakshraj Singh Manav Muthanna, M. Sivasankaran, K. School of Electronics Engineering Vellore Institute of Technology Vellore India School of Computer Science Engineering and Information Systems Vellore Institute of Technology Vellore India

ISBN: (纸本)9798331527549

Road accidents are a significant global issue, with human error causing a substantial number of collisions. This emphasizes the need for advanced collision avoidance systems to improve road safety. In response, this study integrates object detection and distance estimation techniques to enhance vehicle safety. Utilizing recent iterations of the YOLO architecture—YOLOv6n, YOLOv8n, YOLOv9t, and YOLOv10n—the models were trained and evaluated on the KITTI dataset, a benchmark suite for driving applications. This research study provides a comparative analysis of these models, focusing on performance metrics such as precision, mean average precision, recall, and F1 score. A novel single-camera distance estimation method was also implemented to assess object proximity accurately. Among the evaluated models, YOLOv8n achieved the best results, with a F1 score of 0.74 and a mAP@0.5 of 0.782, demonstrating superior detection accuracy and reliability. Its swift convergence and low training losses make it well-suited for real-world use cases. While YOLOv9t and YOLOv10n also performed well, they were slightly less effective. The system enhances its capabilities by accurately estimating distances, which improves collision avoidance and road safety applications. Overall, YOLOv8n is the most efficient and reliable option for developing advanced safety features in vehicle systems. © 2025 IEEE.

关键词： Vehicle safety

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：