Computers & Electrical Engineering最新文献_第4页

C3E: A framework for chart classification and content extraction C3E：图表分类和内容提取框架

IF 4 3区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Computers & Electrical Engineering

Pub Date : 2024-11-22 DOI: 10.1016/j.compeleceng.2024.109861

Muhammad Suhaib Kanroo , Hadia Showkat Kawoosa , Kapil Rana , Puneet Goyal

Incorporating charts into technical documents enhances richness by simplifying complex data representation and improving comprehension. However, automated chart content extraction (CCE) presents a significant challenge within the domain of document analysis and understanding. The CCE problem can be viewed through a series of six sub-tasks: chart classification (CC), text detection and recognition (TDR), text role classification (TRC), axis analysis, legend analysis, and data extraction. Improving these sub-tasks is important for enhancing the effectiveness of CCE. This paper introduces the chart classification and content extraction (C3E) framework, with a primary focus on the first three sub-tasks of CCE: CC, TDR, and TRC. We propose a ChartVision model for the CC, an EfficientNet-based model coupled with a dual-branch architecture incorporating a novel hybrid convolutional and dilated attention module. For text detection and TRC, we introduce a novel CCE method based on YOLOv5, CCE-YOLO, designed for localizing and classifying textual components of varying sizes. Further, for text recognition, we employ a convolutional recurrent neural network with connectionist temporal classification loss. We conducted experimental analysis on benchmark datasets to assess the effectiveness of our approach across each sub-task. Specifically, we evaluated CC, TDR, and TRC methods using the UB-PMC 2020 and UB-PMC 2022 datasets from the ICPR2020 and ICPR2022 CHART-Infographics competitions. The C3E framework achieved notable F1-scores of 94.26%, 92.44%, and 80.64% for CC, TDR, and TRC, respectively on the UB-PMC 2020 dataset and 94.0%, 91.98%, and 84.48% for CC, TDR, and TRC, respectively on the UB-PMC 2022 dataset.

在技术文档中加入图表可以简化复杂的数据表示并提高理解能力，从而增强文档的丰富性。然而，自动图表内容提取（CCE）是文档分析和理解领域的一项重大挑战。CCE 问题可通过一系列六个子任务来看待：图表分类 (CC)、文本检测和识别 (TDR)、文本角色分类 (TRC)、轴分析、图例分析和数据提取。改进这些子任务对于提高 CCE 的有效性非常重要。本文介绍了图表分类和内容提取（C3E）框架，主要侧重于 CCE 的前三个子任务：CC、TDR 和 TRC。我们为 CC 提出了一个 ChartVision 模型，这是一个基于 EfficientNet 的模型，并结合了一个新颖的混合卷积和扩张注意力模块的双分支架构。在文本检测和 TRC 方面，我们引入了一种基于 YOLOv5 的新型 CCE 方法，即 CCE-YOLO，专门用于对不同大小的文本成分进行定位和分类。此外，在文本识别方面，我们采用了具有连接主义时序分类损失的卷积递归神经网络。我们在基准数据集上进行了实验分析，以评估我们的方法在各个子任务中的有效性。具体来说，我们使用来自 ICPR2020 和 ICPR2022 CHART-Infographics 竞赛的 UB-PMC 2020 和 UB-PMC 2022 数据集对 CC、TDR 和 TRC 方法进行了评估。在 UB-PMC 2020 数据集上，CC、TDR 和 TRC 的 F1 分数分别为 94.26%、92.44% 和 80.64%；在 UB-PMC 2022 数据集上，CC、TDR 和 TRC 的 F1 分数分别为 94.0%、91.98% 和 84.48%。

{"title":"C3E: A framework for chart classification and content extraction","authors":"Muhammad Suhaib Kanroo , Hadia Showkat Kawoosa , Kapil Rana , Puneet Goyal","doi":"10.1016/j.compeleceng.2024.109861","DOIUrl":"10.1016/j.compeleceng.2024.109861","url":null,"abstract":"<div><div>Incorporating charts into technical documents enhances richness by simplifying complex data representation and improving comprehension. However, automated chart content extraction (CCE) presents a significant challenge within the domain of document analysis and understanding. The CCE problem can be viewed through a series of six sub-tasks: chart classification (CC), text detection and recognition (TDR), text role classification (TRC), axis analysis, legend analysis, and data extraction. Improving these sub-tasks is important for enhancing the effectiveness of CCE. This paper introduces the chart classification and content extraction (C3E) framework, with a primary focus on the first three sub-tasks of CCE: CC, TDR, and TRC. We propose a ChartVision model for the CC, an EfficientNet-based model coupled with a dual-branch architecture incorporating a novel hybrid convolutional and dilated attention module. For text detection and TRC, we introduce a novel CCE method based on YOLOv5, CCE-YOLO, designed for localizing and classifying textual components of varying sizes. Further, for text recognition, we employ a convolutional recurrent neural network with connectionist temporal classification loss. We conducted experimental analysis on benchmark datasets to assess the effectiveness of our approach across each sub-task. Specifically, we evaluated CC, TDR, and TRC methods using the UB-PMC 2020 and UB-PMC 2022 datasets from the ICPR2020 and ICPR2022 CHART-Infographics competitions. The C3E framework achieved notable F1-scores of 94.26%, 92.44%, and 80.64% for CC, TDR, and TRC, respectively on the UB-PMC 2020 dataset and 94.0%, 91.98%, and 84.48% for CC, TDR, and TRC, respectively on the UB-PMC 2022 dataset.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"121 ","pages":"Article 109861"},"PeriodicalIF":4.0,"publicationDate":"2024-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142701821","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"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

Nonlinear robust integral based actor–critic reinforcement learning control for a perturbed three-wheeled mobile robot with mecanum wheels 基于行为批判强化学习的非线性鲁棒积分控制，适用于带机械轮的扰动三轮移动机器人

IF 4 3区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Computers & Electrical Engineering

Pub Date : 2024-11-22 DOI: 10.1016/j.compeleceng.2024.109870

Phuong Nam Dao, Minh Hiep Phung

In this article, a novel Robust Integral of the Sign of the Error (RISE)-based Actor/Critic reinforcement learning control structure is established, which addresses the trajectory tracking control problem, optimality performance and observer effectiveness of a three mecanum wheeled mobile robot to be subject to slipping effect. The Actor–Critic Reinforcement Learning algorithm with a discount factor is introduced in integration with the Nonlinear RISE feedback term, which is designated to eliminate the dynamic uncertainties/disturbances from the affine nominal system. On the other hand, the persistence of excitation (PE) condition can be ignored due to the presence of RISE term. Stability analyses in two proposed theorems demonstrate all the signals in the closed-loop system and learning weights would be Uniformly Ultimate Boundedness (UUB) and the consideration of the system under the impact of RISE that can promote the tracking effectiveness. In conclusion, simulation results are shown in conjunction with the comparison to illustrate the powerful capability as well as the economy in control resources of the proposed algorithm.

本文建立了一种新颖的基于误差符号稳健积分（RISE）的演员/批判强化学习控制结构，解决了受滑移效应影响的三轴轮式移动机器人的轨迹跟踪控制问题、最优性能和观测器有效性问题。将带有贴现因子的行动者批判强化学习算法与非线性 RISE 反馈项结合起来，用于消除仿射标称系统的动态不确定性/扰动。另一方面，由于 RISE 项的存在，激励持续性（PE）条件可被忽略。所提出的两个定理中的稳定性分析表明，闭环系统中的所有信号和学习权重都是均匀终极有界的（UUB），而且考虑到系统在 RISE 的影响下可以提高跟踪效果。最后，仿真结果与比较结果相结合，说明了所提算法的强大能力和控制资源的经济性。

{"title":"Nonlinear robust integral based actor–critic reinforcement learning control for a perturbed three-wheeled mobile robot with mecanum wheels","authors":"Phuong Nam Dao, Minh Hiep Phung","doi":"10.1016/j.compeleceng.2024.109870","DOIUrl":"10.1016/j.compeleceng.2024.109870","url":null,"abstract":"<div><div>In this article, a novel Robust Integral of the Sign of the Error (RISE)-based Actor/Critic reinforcement learning control structure is established, which addresses the trajectory tracking control problem, optimality performance and observer effectiveness of a three mecanum wheeled mobile robot to be subject to slipping effect. The Actor–Critic Reinforcement Learning algorithm with a discount factor is introduced in integration with the Nonlinear RISE feedback term, which is designated to eliminate the dynamic uncertainties/disturbances from the affine nominal system. On the other hand, the persistence of excitation (PE) condition can be ignored due to the presence of RISE term. Stability analyses in two proposed theorems demonstrate all the signals in the closed-loop system and learning weights would be Uniformly Ultimate Boundedness (UUB) and the consideration of the system under the impact of RISE that can promote the tracking effectiveness. In conclusion, simulation results are shown in conjunction with the comparison to illustrate the powerful capability as well as the economy in control resources of the proposed algorithm.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"121 ","pages":"Article 109870"},"PeriodicalIF":4.0,"publicationDate":"2024-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142701819","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"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 monocular three-dimensional object detection model based on uncertainty-guided depth combination for autonomous driving 基于不确定性引导深度组合的单目三维物体检测模型，用于自动驾驶

IF 4 3区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Computers & Electrical Engineering

Pub Date : 2024-11-21 DOI: 10.1016/j.compeleceng.2024.109864

Xin Zhou , Xiaolong Xu

Three-Dimensional (3D) object detection is a crucial task for enhancing safety and efficiency in autonomous driving. However, estimating depth from monocular images remains a challenging task. Most existing monocular 3D object detection methods rely on additional auxiliary data sources to compensate for the lack of spatial information in monocular images. Nevertheless, these methods bring substantial computational overhead and time-consuming preprocessing steps. To address this issue, we propose a novel depth estimation method for monocular images that does not rely on any auxiliary information. Leveraging both the texture and geometric cues of detected objects, our method generates two depth estimates for each object based on the extracted Region of Interest (RoI) features: a direct depth estimate and a height-based depth estimate with uncertainty modeling. Our model dynamically assigns weights to these depth estimates based on their respective uncertainties and combines them to obtain the final depth. During the training process, the model assigns higher weights to depth branches with higher uncertainties, as these estimates exhibit greater tolerance to errors. As the combined depth network introduces increased complexity, we utilize Group Normalization (GN) to better capture spatial information in the prediction branch outputs. Furthermore, we leverage the Two-Dimensional (2D) information of objects to predict the residual in the 2D center after downsampling, aiding in the regression of 3D center. On the KITTI benchmark, our model achieves an average precision (AP) of 16.65 % and 23.19 % on 3D and bird's-eye view (BEV) detection for the moderate category, surpassing the state-of-the-art (SOTA) models in each category.

三维（3D）物体检测是提高自动驾驶安全性和效率的关键任务。然而，从单目图像中估计深度仍然是一项具有挑战性的任务。现有的大多数单目三维物体检测方法都依赖于额外的辅助数据源，以弥补单目图像中空间信息的不足。然而，这些方法带来了大量的计算开销和耗时的预处理步骤。为了解决这个问题，我们提出了一种不依赖任何辅助信息的新型单目图像深度估计方法。利用检测到的物体的纹理和几何线索，我们的方法可根据提取的感兴趣区域（RoI）特征为每个物体生成两种深度估计值：直接深度估计值和基于高度的不确定性建模深度估计值。我们的模型会根据这些深度估计值各自的不确定性为其动态分配权重，并将它们结合起来以获得最终深度。在训练过程中，模型会为不确定性较高的深度分支分配更高的权重，因为这些估计值对误差的容忍度更高。由于组合深度网络带来了更高的复杂性，我们利用组归一化（GN）技术在预测分支输出中更好地捕捉空间信息。此外，我们还利用物体的二维（2D）信息来预测下采样后二维中心的残差，从而帮助回归三维中心。在 KITTI 基准测试中，我们的模型在中度类别的三维和鸟瞰 (BEV) 检测中分别达到了 16.65% 和 23.19% 的平均精度 (AP)，在每个类别中都超过了最先进的模型 (SOTA)。

{"title":"A monocular three-dimensional object detection model based on uncertainty-guided depth combination for autonomous driving","authors":"Xin Zhou , Xiaolong Xu","doi":"10.1016/j.compeleceng.2024.109864","DOIUrl":"10.1016/j.compeleceng.2024.109864","url":null,"abstract":"<div><div>Three-Dimensional (3D) object detection is a crucial task for enhancing safety and efficiency in autonomous driving. However, estimating depth from monocular images remains a challenging task. Most existing monocular 3D object detection methods rely on additional auxiliary data sources to compensate for the lack of spatial information in monocular images. Nevertheless, these methods bring substantial computational overhead and time-consuming preprocessing steps. To address this issue, we propose a novel depth estimation method for monocular images that does not rely on any auxiliary information. Leveraging both the texture and geometric cues of detected objects, our method generates two depth estimates for each object based on the extracted Region of Interest (RoI) features: a direct depth estimate and a height-based depth estimate with uncertainty modeling. Our model dynamically assigns weights to these depth estimates based on their respective uncertainties and combines them to obtain the final depth. During the training process, the model assigns higher weights to depth branches with higher uncertainties, as these estimates exhibit greater tolerance to errors. As the combined depth network introduces increased complexity, we utilize Group Normalization (GN) to better capture spatial information in the prediction branch outputs. Furthermore, we leverage the Two-Dimensional (2D) information of objects to predict the residual in the 2D center after downsampling, aiding in the regression of 3D center. On the KITTI benchmark, our model achieves an average precision (AP) of 16.65 % and 23.19 % on 3D and bird's-eye view (BEV) detection for the moderate category, surpassing the state-of-the-art (SOTA) models in each category.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"120 ","pages":"Article 109864"},"PeriodicalIF":4.0,"publicationDate":"2024-11-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142698895","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"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 smart contract solution for transparent auctions on permissioned blockchain platform 许可区块链平台上透明拍卖的智能合约解决方案

IF 4 3区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Computers & Electrical Engineering

Pub Date : 2024-11-21 DOI: 10.1016/j.compeleceng.2024.109859

Sujata Swain , Vikas Chouhan

The rapid and advanced development of the Internet and Technology in recent years has led to the increased popularity of online electronic auctioning (e-auction) systems like eBay. These e-auctions are becoming increasingly essential to the global economy and hold a promising future as they increase user participation and provide advantages such as time-saving, low cost, and ubiquity (not limited to geographic location). However, existing real-time e-auction systems are centralized and rely on intermediaries, leading to issues such as lack of transparency, corruption, security risks, interoperability, and low trust from bidders and auctioneers. To address these issues, we present a permissioned blockchain-based transparent auction system that allows participants to participate in the auction and bidder organization in a single decentralized platform. Auctioneer announces the auction then bidders submit an individual bid for the auction during the auction period. In this paper, we present the implementation of smart contracts over the Hyperledger Fabric platform, conduct testbeds using the caliper benchmark tool, and report the results.

近年来，互联网和技术的迅猛发展使 eBay 等在线电子拍卖系统日益普及。这些电子拍卖对全球经济越来越重要，而且前景广阔，因为它们提高了用户参与度，并具有省时、低成本和无处不在（不受地理位置限制）等优势。然而，现有的实时电子拍卖系统都是集中式的，依赖于中介机构，导致了缺乏透明度、腐败、安全风险、互操作性以及投标人和拍卖人信任度低等问题。为了解决这些问题，我们提出了一种基于许可的区块链透明拍卖系统，它允许参与者在一个单一的去中心化平台上参与拍卖和组织竞标。拍卖人宣布拍卖，然后投标人在拍卖期间提交个人投标。在本文中，我们介绍了在 Hyperledger Fabric 平台上实现智能合约的情况，使用 caliper 基准工具进行了测试，并报告了测试结果。

{"title":"A smart contract solution for transparent auctions on permissioned blockchain platform","authors":"Sujata Swain , Vikas Chouhan","doi":"10.1016/j.compeleceng.2024.109859","DOIUrl":"10.1016/j.compeleceng.2024.109859","url":null,"abstract":"<div><div>The rapid and advanced development of the Internet and Technology in recent years has led to the increased popularity of online electronic auctioning (e-auction) systems like eBay. These e-auctions are becoming increasingly essential to the global economy and hold a promising future as they increase user participation and provide advantages such as time-saving, low cost, and ubiquity (not limited to geographic location). However, existing real-time e-auction systems are centralized and rely on intermediaries, leading to issues such as lack of transparency, corruption, security risks, interoperability, and low trust from bidders and auctioneers. To address these issues, we present a permissioned blockchain-based transparent auction system that allows participants to participate in the auction and bidder organization in a single decentralized platform. Auctioneer announces the auction then bidders submit an individual bid for the auction during the auction period. In this paper, we present the implementation of smart contracts over the Hyperledger Fabric platform, conduct testbeds using the caliper benchmark tool, and report the results.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"121 ","pages":"Article 109859"},"PeriodicalIF":4.0,"publicationDate":"2024-11-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142701818","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"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

Detection and restoration of abnormal band data in photometric images 测光图像中异常波段数据的检测和修复

IF 4 3区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Computers & Electrical Engineering

Pub Date : 2024-11-21 DOI: 10.1016/j.compeleceng.2024.109871

Guoqing Wang , Bo Qiu , Ali Luo , Xiao Kong , Zhiren Pan , Qi Li , Fuji Ren , Guanlong Cao

Addressing the issue of abnormal band data processing in photometric surveys is imperative. Restoring of abnormal band data not only salvages a significant amount of existing astronomical observation data, but also has profound implications on the data processing of new optical telescopes in the future. This paper firstly designs Band Data MogaNet(BDMogaNet) classification model for normal or abnormal band data, which can automatically identify abnormal data. Then, for the restoration of abnormal band data, Global–Local Recursive Generalization(GLRG) restoration network is designed. The experiment used the SDSS image library, and the results proved that the classification accuracy of normal band data and abnormal band data using BDMogaNet reached 99.2% in the training set and 98.0% in the validation set, which had a better classification comparing to some newest methods. Moreover, PSNR of restoring abnormal band data using GLRG reached 33.96 dB, SSIM reached 0.73, and CM reached 6.09, which are all better compared to some newest methods.

解决光度测量中的异常波段数据处理问题势在必行。恢复异常波段数据不仅可以挽救大量现有的天文观测数据，而且对未来新型光学望远镜的数据处理具有深远的影响。本文首先设计了波段数据 MogaNet（BDMogaNet）分类模型，可自动识别正常或异常波段数据。然后，针对异常波段数据的恢复，设计了全局-局部递归泛化（GLRG）恢复网络。实验使用了 SDSS 图像库，结果证明，使用 BDMogaNet 对正常波段数据和异常波段数据的分类准确率在训练集上达到了 99.2%，在验证集上达到了 98.0%，与一些最新方法相比，分类效果更好。此外，使用 GLRG 还原异常波段数据的 PSNR 达到 33.96 dB，SSIM 达到 0.73，CM 达到 6.09，均优于一些最新方法。

{"title":"Detection and restoration of abnormal band data in photometric images","authors":"Guoqing Wang , Bo Qiu , Ali Luo , Xiao Kong , Zhiren Pan , Qi Li , Fuji Ren , Guanlong Cao","doi":"10.1016/j.compeleceng.2024.109871","DOIUrl":"10.1016/j.compeleceng.2024.109871","url":null,"abstract":"<div><div>Addressing the issue of abnormal band data processing in photometric surveys is imperative. Restoring of abnormal band data not only salvages a significant amount of existing astronomical observation data, but also has profound implications on the data processing of new optical telescopes in the future. This paper firstly designs Band Data MogaNet(BDMogaNet) classification model for normal or abnormal band data, which can automatically identify abnormal data. Then, for the restoration of abnormal band data, Global–Local Recursive Generalization(GLRG) restoration network is designed. The experiment used the SDSS image library, and the results proved that the classification accuracy of normal band data and abnormal band data using BDMogaNet reached 99.2% in the training set and 98.0% in the validation set, which had a better classification comparing to some newest methods. Moreover, PSNR of restoring abnormal band data using GLRG reached 33.96 dB, SSIM reached 0.73, and CM reached 6.09, which are all better compared to some newest methods.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"121 ","pages":"Article 109871"},"PeriodicalIF":4.0,"publicationDate":"2024-11-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142701817","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"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 machine learning-based approach for maximizing system profit in a power system by imbalance price curtailment 基于机器学习的电力系统不平衡价格削减系统利润最大化方法

IF 4 3区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Computers & Electrical Engineering

Pub Date : 2024-11-20 DOI: 10.1016/j.compeleceng.2024.109874

Shreya Shree Das , Priyanka Singh , Jayendra Kumar , Subhojit Dawn , Anumoy Ghosh

The integration of wind farms into the power grid is difficult due to unpredictable wind speed fluctuations. This variation has an impact on power generation profitability, demanding effective forecasting to lessen pricing risks. A novel optimization approach is proposed in this paper to expand social welfare and profitability while increasing revenue for power generators. This method is crucial for avoiding financial risks related to variable wind patterns. Narrowing the gap between anticipated and actual wind speeds (WS_AN, WS_AC) can lessen the negative impact of imbalanced prices on profitability. This reduction is necessary to enhance the economic performance of the power system. The paper endorses the use of machine learning (ML) techniques, specifically Long Short-Term Memory (LSTM) and Random Forest (RF) methods, to precisely anticipate wind speed. These models serve as analytical tools for enlightening decision-making and resource allocation in wind energy generation. According to the study, pricing imbalances have a major impression on profit calculations in deregulated systems. The empirical data show that effective forecasting can expand financial outcomes for energy companies, reducing risks and maximizing revenue. Finally, the empirical results highlight the significance of accurate wind speed forecasts and the use of advanced optimization approaches in growing the profitability and efficiency of renewable energy-dependent power systems. These findings offer a strong foundation for further research and use of machine learning techniques in the energy sector. The optimization model was accomplished with modified IEEE 14 bus test systems in this work.

由于风速波动难以预测，将风力发电场并入电网非常困难。这种变化会影响发电利润，因此需要进行有效预测以降低定价风险。本文提出了一种新颖的优化方法，以扩大社会福利和盈利能力，同时增加发电企业的收入。这种方法对于避免与多变风力模式相关的财务风险至关重要。缩小预期风速与实际风速之间的差距（WSAN、WSAC）可以减少不平衡价格对盈利能力的负面影响。要提高电力系统的经济效益，就必须缩小这种差距。本文赞同使用机器学习 (ML) 技术，特别是长短期记忆 (LSTM) 和随机森林 (RF) 方法来精确预测风速。这些模型可作为风能发电决策和资源分配的分析工具。研究表明，定价失衡对放松管制系统中的利润计算有重大影响。实证数据表明，有效的预测可以扩大能源公司的财务成果，降低风险并实现收益最大化。最后，实证结果突出表明，准确的风速预测和先进优化方法的使用对于提高依赖可再生能源的电力系统的盈利能力和效率具有重要意义。这些发现为在能源领域进一步研究和使用机器学习技术奠定了坚实的基础。在这项工作中，优化模型是通过修改后的 IEEE 14 总线测试系统完成的。

{"title":"A machine learning-based approach for maximizing system profit in a power system by imbalance price curtailment","authors":"Shreya Shree Das , Priyanka Singh , Jayendra Kumar , Subhojit Dawn , Anumoy Ghosh","doi":"10.1016/j.compeleceng.2024.109874","DOIUrl":"10.1016/j.compeleceng.2024.109874","url":null,"abstract":"<div><div>The integration of wind farms into the power grid is difficult due to unpredictable wind speed fluctuations. This variation has an impact on power generation profitability, demanding effective forecasting to lessen pricing risks. A novel optimization approach is proposed in this paper to expand social welfare and profitability while increasing revenue for power generators. This method is crucial for avoiding financial risks related to variable wind patterns. Narrowing the gap between anticipated and actual wind speeds (WS<sub>AN</sub>, WS<sub>AC</sub>) can lessen the negative impact of imbalanced prices on profitability. This reduction is necessary to enhance the economic performance of the power system. The paper endorses the use of machine learning (ML) techniques, specifically Long Short-Term Memory (LSTM) and Random Forest (RF) methods, to precisely anticipate wind speed. These models serve as analytical tools for enlightening decision-making and resource allocation in wind energy generation. According to the study, pricing imbalances have a major impression on profit calculations in deregulated systems. The empirical data show that effective forecasting can expand financial outcomes for energy companies, reducing risks and maximizing revenue. Finally, the empirical results highlight the significance of accurate wind speed forecasts and the use of advanced optimization approaches in growing the profitability and efficiency of renewable energy-dependent power systems. These findings offer a strong foundation for further research and use of machine learning techniques in the energy sector. The optimization model was accomplished with modified IEEE 14 bus test systems in this work.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"121 ","pages":"Article 109874"},"PeriodicalIF":4.0,"publicationDate":"2024-11-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142701816","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"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

Identifying social concerns in virtual reality technology through text mining and large language models, and prioritizing them with the fuzzy hierarchized analytic network process 通过文本挖掘和大型语言模型识别虚拟现实技术中的社会问题，并利用模糊分层分析网络流程对其进行优先排序

IF 4 3区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Computers & Electrical Engineering

Pub Date : 2024-11-20 DOI: 10.1016/j.compeleceng.2024.109770

Esmaeil Rezaei , Behzad Mosallanezhad

Virtual reality technology has rapidly gained popularity as an entertainment medium, drawing interest from diverse age groups. However, its widespread adoption depends on effectively addressing public concerns and achieving market acceptance. While some studies have acknowledged these concerns, a significant gap persists in comprehensive research that incorporates both individual and expert perspectives. Consequently, certain underlying social issues related to virtual reality systems remain unexplored and unprioritized. To address this gap, this paper proposes a methodology that utilizes Latent Semantic Analysis (LSA) to identify and assess social concerns from various sources, including user perspectives. Large Language Models (LLMs) assist in retrieving relevant chunks of articles during analysis, enhancing data quality. Furthermore, we introduce a novel decision-making tool, the Hierarchized Analytic Network Process (HANP) and its fuzzy form, to effectively rank these concerns. This approach addresses a limitation of the traditional Analytic Network Process (ANP), which can overemphasize dependent attributes, potentially leading to zero-weighted, less important attributes and making comparisons impossible. By prioritizing social concerns based on their significance, our approach aims to facilitate broader social acceptance of virtual reality technologies among the general public. To further demonstrate the advantages of our proposed approach, the results obtained from F-HANP (in situations where fuzzy judgments are available) and HANP are compared with other popular decision-making methods.

虚拟现实技术作为一种娱乐媒介迅速普及，吸引了不同年龄群体的兴趣。然而，虚拟现实技术能否得到广泛应用，取决于能否有效解决公众关注的问题并获得市场认可。虽然一些研究已经认识到了这些问题，但在结合个人和专家观点的综合研究方面仍然存在巨大差距。因此，与虚拟现实系统相关的某些潜在社会问题仍未得到探讨和重视。为了弥补这一不足，本文提出了一种方法，利用潜在语义分析（LSA）从各种来源（包括用户视角）识别和评估社会问题。大型语言模型（LLM）可在分析过程中协助检索相关的文章块，从而提高数据质量。此外，我们还引入了一种新颖的决策工具--分层分析网络流程（HANP）及其模糊形式，以有效地对这些关注点进行排序。这种方法解决了传统分析网络流程（ANP）的一个局限性，即它可能会过度强调从属属性，从而可能导致零权重、不太重要的属性，并使比较变得不可能。我们的方法根据社会关注点的重要性对其进行优先排序，旨在促进社会大众更广泛地接受虚拟现实技术。为了进一步证明我们提出的方法的优势，我们将 F-HANP（在有模糊判断的情况下）和 HANP 得出的结果与其他流行的决策方法进行了比较。

{"title":"Identifying social concerns in virtual reality technology through text mining and large language models, and prioritizing them with the fuzzy hierarchized analytic network process","authors":"Esmaeil Rezaei , Behzad Mosallanezhad","doi":"10.1016/j.compeleceng.2024.109770","DOIUrl":"10.1016/j.compeleceng.2024.109770","url":null,"abstract":"<div><div>Virtual reality technology has rapidly gained popularity as an entertainment medium, drawing interest from diverse age groups. However, its widespread adoption depends on effectively addressing public concerns and achieving market acceptance. While some studies have acknowledged these concerns, a significant gap persists in comprehensive research that incorporates both individual and expert perspectives. Consequently, certain underlying social issues related to virtual reality systems remain unexplored and unprioritized. To address this gap, this paper proposes a methodology that utilizes Latent Semantic Analysis (LSA) to identify and assess social concerns from various sources, including user perspectives. Large Language Models (LLMs) assist in retrieving relevant chunks of articles during analysis, enhancing data quality. Furthermore, we introduce a novel decision-making tool, the Hierarchized Analytic Network Process (HANP) and its fuzzy form, to effectively rank these concerns. This approach addresses a limitation of the traditional Analytic Network Process (ANP), which can overemphasize dependent attributes, potentially leading to zero-weighted, less important attributes and making comparisons impossible. By prioritizing social concerns based on their significance, our approach aims to facilitate broader social acceptance of virtual reality technologies among the general public. To further demonstrate the advantages of our proposed approach, the results obtained from F-HANP (in situations where fuzzy judgments are available) and HANP are compared with other popular decision-making methods.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"120 ","pages":"Article 109770"},"PeriodicalIF":4.0,"publicationDate":"2024-11-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142699938","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"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

Artificial neural network-based virtual synchronous generator for frequency stability improving of grid integrating distributed generators 基于人工神经网络的虚拟同步发电机用于提高分布式发电机并网发电的频率稳定性

IF 4 3区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Computers & Electrical Engineering

Pub Date : 2024-11-20 DOI: 10.1016/j.compeleceng.2024.109877

Abderrahmane Smahi , Salim Makhloufi

The integration of renewable energy sources (RESs) is becoming increasingly prevalent in contemporary power grids. RESs, including distributed generators (DGs), utilize power electronics converters to interface with the grid, contributing to a reduction in grid inertia and an increase in vulnerability to stability issues. This shift has led to a gradual displacement of the traditional role of synchronous generators (SGs) in providing frequency regulation, with power electronics converters such as inverters taking on a more prominent role. Virtual synchronous generators (VSGs) or virtual synchronous machines (VSMs) offer a solution by emulating SG behavior in power electronics converters. However, these techniques encounter limitations in mathematical calculations and precision. This article proposes an artificial intelligent based VSM controller (AIVSM) designed to overcome these limitations. The AIVSM system leverages artificial neural networks (ANNs) to emulate real SGs. The ANN is trained using a substantial dataset derived from a SG of a diesel generator. Simulation results demonstrate the performance superiority of the AIVSM when compared to a conventional proportional integral (PI) VSM controller and an adaptive VSM controller.

在当代电网中，可再生能源（RES）的整合正变得越来越普遍。包括分布式发电机 (DG) 在内的可再生能源利用电力电子变流器与电网连接，从而降低了电网惯性，并增加了对稳定性问题的脆弱性。这种转变导致同步发电机（SG）在提供频率调节方面的传统作用逐渐被取代，变频器等电力电子变流器的作用更加突出。虚拟同步发电机 (VSG) 或虚拟同步机 (VSM) 通过在电力电子转换器中模拟同步发电机的行为提供了一种解决方案。然而，这些技术在数学计算和精度方面存在局限性。本文提出了一种基于人工智能的 VSM 控制器 (AIVSM)，旨在克服这些限制。AIVSM 系统利用人工神经网络 (ANN) 来模拟真实的 SG。人工神经网络是利用柴油发电机 SG 的大量数据集进行训练的。仿真结果表明，与传统的比例积分 (PI) VSM 控制器和自适应 VSM 控制器相比，AIVSM 性能更优越。

{"title":"Artificial neural network-based virtual synchronous generator for frequency stability improving of grid integrating distributed generators","authors":"Abderrahmane Smahi , Salim Makhloufi","doi":"10.1016/j.compeleceng.2024.109877","DOIUrl":"10.1016/j.compeleceng.2024.109877","url":null,"abstract":"<div><div>The integration of renewable energy sources (RESs) is becoming increasingly prevalent in contemporary power grids. RESs, including distributed generators (DGs), utilize power electronics converters to interface with the grid, contributing to a reduction in grid inertia and an increase in vulnerability to stability issues. This shift has led to a gradual displacement of the traditional role of synchronous generators (SGs) in providing frequency regulation, with power electronics converters such as inverters taking on a more prominent role. Virtual synchronous generators (VSGs) or virtual synchronous machines (VSMs) offer a solution by emulating SG behavior in power electronics converters. However, these techniques encounter limitations in mathematical calculations and precision. This article proposes an artificial intelligent based VSM controller (AIVSM) designed to overcome these limitations. The AIVSM system leverages artificial neural networks (ANNs) to emulate real SGs. The ANN is trained using a substantial dataset derived from a SG of a diesel generator. Simulation results demonstrate the performance superiority of the AIVSM when compared to a conventional proportional integral (PI) VSM controller and an adaptive VSM controller.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"120 ","pages":"Article 109877"},"PeriodicalIF":4.0,"publicationDate":"2024-11-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142699500","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"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

High frequency domain enhancement and channel attention module for multi-view stereo 用于多视角立体声的高频域增强和通道关注模块

IF 4 3区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Computers & Electrical Engineering

Pub Date : 2024-11-20 DOI: 10.1016/j.compeleceng.2024.109855

Yongjuan Yang , Jie Cao , Hong Zhao , Zhaobin Chang , Weijie Wang

Multi-view stereo based on deep learning is increasingly popular as a method for 3D reconstruction. Existing methods have made significant advancements in pixel-level depth estimation. However, challenges such as occlusions and non-Lambertian surfaces in images hinder accurate confidence estimation. Moreover, cost volume regularization often results in excessive smoothing at object boundaries. To tackle these challenges, we propose integrating the High Frequency Information Compensator and 3D Channel Attention Module into the Multi-View Stereo Network, termed HFCA-MVS. Firstly, in the feature volume aggregation stage, we introduce a high-frequency information compensator module to enhance the correlation between 2D semantics and 3D space. Subsequently, in the cost volume regularization stage, a 3D channel attention module is introduced to enhance the representation of channel features by capturing relationships among different channels. Lastly, the 3DCNN network employs the GELU activation function to boost the activation response and mitigate excessive object boundary smoothing. HFCA-MVS demonstrates competitive performance in 3D reconstruction across three benchmark datasets: DTU, BlendMVS, and Tanks&Temples. Particularly, compared to CasMVSNet, MVSTER, and Geo-MVSNet on the DTU benchmark, HFCA-MVS achieves a relative improvement in completeness of 33%, 6.5%, and 0.4%, respectively, and an enhancement in overall performance of 15% and 4.2% compared to CasMVSNet and MVSTER. Furthermore, our model yields comparable reconstruction results to existing models on the Tanks&Temples dataset.

作为一种三维重建方法，基于深度学习的多视角立体技术越来越受欢迎。现有方法在像素级深度估计方面取得了显著进步。然而，图像中的遮挡和非朗伯表面等挑战阻碍了准确的置信度估计。此外，成本体积正则化往往会导致物体边界过度平滑。为了应对这些挑战，我们建议将高频信息补偿器和三维通道注意模块集成到多视图立体网络中，称为 HFCA-MVS。首先，在特征卷聚合阶段，我们引入了高频信息补偿器模块，以增强二维语义与三维空间之间的相关性。随后，在代价卷正则化阶段，我们引入了三维信道关注模块，通过捕捉不同信道之间的关系来增强信道特征的表示。最后，3DCNN 网络采用 GELU 激活函数来增强激活响应，并减少过度的对象边界平滑。HFCA-MVS 在三个基准数据集的三维重建中表现出了极具竞争力的性能：DTU、BlendMVS 和 Tanks&Temples。特别是在 DTU 基准数据集上，与 CasMVSNet、MVSTER 和 Geo-MVSNet 相比，HFCA-MVS 的完整性分别提高了 33%、6.5% 和 0.4%，总体性能比 CasMVSNet 和 MVSTER 分别提高了 15% 和 4.2%。此外，我们的模型在 Tanks&Temples 数据集上获得了与现有模型相当的重建结果。

{"title":"High frequency domain enhancement and channel attention module for multi-view stereo","authors":"Yongjuan Yang , Jie Cao , Hong Zhao , Zhaobin Chang , Weijie Wang","doi":"10.1016/j.compeleceng.2024.109855","DOIUrl":"10.1016/j.compeleceng.2024.109855","url":null,"abstract":"<div><div>Multi-view stereo based on deep learning is increasingly popular as a method for 3D reconstruction. Existing methods have made significant advancements in pixel-level depth estimation. However, challenges such as occlusions and non-Lambertian surfaces in images hinder accurate confidence estimation. Moreover, cost volume regularization often results in excessive smoothing at object boundaries. To tackle these challenges, we propose integrating the High Frequency Information Compensator and 3D Channel Attention Module into the Multi-View Stereo Network, termed HFCA-MVS. Firstly, in the feature volume aggregation stage, we introduce a high-frequency information compensator module to enhance the correlation between 2D semantics and 3D space. Subsequently, in the cost volume regularization stage, a 3D channel attention module is introduced to enhance the representation of channel features by capturing relationships among different channels. Lastly, the 3DCNN network employs the GELU activation function to boost the activation response and mitigate excessive object boundary smoothing. HFCA-MVS demonstrates competitive performance in 3D reconstruction across three benchmark datasets: DTU, BlendMVS, and Tanks&Temples. Particularly, compared to CasMVSNet, MVSTER, and Geo-MVSNet on the DTU benchmark, HFCA-MVS achieves a relative improvement in completeness of 33%, 6.5%, and 0.4%, respectively, and an enhancement in overall performance of 15% and 4.2% compared to CasMVSNet and MVSTER. Furthermore, our model yields comparable reconstruction results to existing models on the Tanks&Temples dataset.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"121 ","pages":"Article 109855"},"PeriodicalIF":4.0,"publicationDate":"2024-11-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142701814","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"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

RCTrans-Net: A spatiotemporal model for fast-time human detection behind walls using ultrawideband radar RCTrans-Net：利用超宽带雷达快速探测墙后人体的时空模型

IF 4 3区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Computers & Electrical Engineering

Pub Date : 2024-11-20 DOI: 10.1016/j.compeleceng.2024.109873

Cries Avian , Jenq-Shiou Leu , Hang Song , Jun-ichi Takada , Nur Achmad Sulistyo Putro , Muhammad Izzuddin Mahali , Setya Widyawan Prakosa

Ultrawideband (UWB) radar systems are becoming increasingly popular for detecting human presence, even through walls. Recent advancements in signal processing use deep learning techniques, which are known for their accuracy. While earlier methods focused on spatial information using Convolutional Neural Networks (CNNs), newer research highlights the importance of temporal information, such as how data peaks shift over time. This study introduces RCTrans-Net, a deep-learning architecture that combines RCNet (a Residual CNN) for spatial features with TransNet (a Transformer) for temporal features. This fusion improves human presence classification in fast-time signal processing. Tested under various conditions—different materials, body orientations, ranges, and radar heights—RCTrans-Net achieved high performance with F1-scores of 0.997±0.000 for static, 0.967±0.004 for dynamic, and 0.978±0.001 for combined scenarios. The architecture outperforms previous methods and offers real-time processing with an inference time of about one millisecond.

超宽带（UWB）雷达系统在探测人的存在方面越来越受欢迎，甚至可以隔墙探测。信号处理领域的最新进展采用了以准确性著称的深度学习技术。早期的方法侧重于使用卷积神经网络（CNN）来处理空间信息，而最新的研究则强调了时间信息的重要性，例如数据峰值如何随时间变化。本研究介绍了 RCTrans-Net，这是一种深度学习架构，它将用于空间特征的 RCNet（残差 CNN）与用于时间特征的 TransNet（变换器）结合在一起。这种融合改进了快速时间信号处理中的人类存在分类。在各种条件下（不同材料、身体方向、射程和雷达高度）进行测试后，RCTrans-Net 取得了很高的性能，静态 F1 分数为 0.997±0.000，动态为 0.967±0.004，综合场景为 0.978±0.001。该架构优于之前的方法，并能提供实时处理，推理时间约为一毫秒。

{"title":"RCTrans-Net: A spatiotemporal model for fast-time human detection behind walls using ultrawideband radar","authors":"Cries Avian , Jenq-Shiou Leu , Hang Song , Jun-ichi Takada , Nur Achmad Sulistyo Putro , Muhammad Izzuddin Mahali , Setya Widyawan Prakosa","doi":"10.1016/j.compeleceng.2024.109873","DOIUrl":"10.1016/j.compeleceng.2024.109873","url":null,"abstract":"<div><div>Ultrawideband (UWB) radar systems are becoming increasingly popular for detecting human presence, even through walls. Recent advancements in signal processing use deep learning techniques, which are known for their accuracy. While earlier methods focused on spatial information using Convolutional Neural Networks (CNNs), newer research highlights the importance of temporal information, such as how data peaks shift over time. This study introduces RCTrans-Net, a deep-learning architecture that combines RCNet (a Residual CNN) for spatial features with TransNet (a Transformer) for temporal features. This fusion improves human presence classification in fast-time signal processing. Tested under various conditions—different materials, body orientations, ranges, and radar heights—RCTrans-Net achieved high performance with F1-scores of 0.997±0.000 for static, 0.967±0.004 for dynamic, and 0.978±0.001 for combined scenarios. The architecture outperforms previous methods and offers real-time processing with an inference time of about one millisecond.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"120 ","pages":"Article 109873"},"PeriodicalIF":4.0,"publicationDate":"2024-11-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142699502","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"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