Pub Date : 2025-11-07DOI: 10.1109/TLA.2025.11231220
Krishna Velmajala;Srinivasa Rao Sandepudi
This article proposes high step-up dual-switch Luo non-isolated DC-DC converter with fault-tolerant capability for critical load applications. The proposed converter constitutes more advantages, which including increased voltage gains with a reduced duty cycle, common grounding between the source and load, low component count and lower voltage and current stress. In addition, it offers reconfiguration capability and decreases the power handling capability by devices, thereby enhancing overall converter efficiency. The performance characteristics of the proposed converter are analyzed in continuous current mode (CCM), with a comprehensive discussion of its features. The proposed converter experimental results are validated at 400 W output power for operational effectiveness and feasibility.
{"title":"High Step-up Dual-switch Luo Non-isolated DC-DC Converter with Fault-tolerant Capability for Critical Load Applications","authors":"Krishna Velmajala;Srinivasa Rao Sandepudi","doi":"10.1109/TLA.2025.11231220","DOIUrl":"https://doi.org/10.1109/TLA.2025.11231220","url":null,"abstract":"This article proposes high step-up dual-switch Luo non-isolated DC-DC converter with fault-tolerant capability for critical load applications. The proposed converter constitutes more advantages, which including increased voltage gains with a reduced duty cycle, common grounding between the source and load, low component count and lower voltage and current stress. In addition, it offers reconfiguration capability and decreases the power handling capability by devices, thereby enhancing overall converter efficiency. The performance characteristics of the proposed converter are analyzed in continuous current mode (CCM), with a comprehensive discussion of its features. The proposed converter experimental results are validated at 400 W output power for operational effectiveness and feasibility.","PeriodicalId":55024,"journal":{"name":"IEEE Latin America Transactions","volume":"23 12","pages":"1271-1283"},"PeriodicalIF":1.3,"publicationDate":"2025-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11231220","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145455851","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-07DOI: 10.1109/TLA.2025.11231218
Ramasamy R;Rajavel V;Rachit Jain
This study investigates the design and performance of millimeter-wave (mm-Wave) Multiple-Input Multiple-Output (MIMO) antennas for fifth-generation (5G) applications, with a particular focus on the consequences of incorporating a ring resonator within the antenna system. This study compares two design variations, one with a ring resonator, and one without to assess their impact on enhancing the antenna's performance characteristics. The research employs five machine learning algorithms, namely, Decision Tree, Random Forest, K-Nearest Neighbors (KNN), XG-Boost, and Gradient Boosting Regressor (GBR), to estimate return loss. Among these, the Random Forest algorithm demonstrates superior performance in terms of accuracy, Mean Squared Error (MSE), Mean Absolute Percentage Error (MAPE), Mean Absolute Error (MAE), and R-squared metrics. The proposed MIMO antenna system shows better performance in Envelope Correlation Coefficient (ECC), Diversity Gain (DG), Channel Capacity Loss (CCL) and Total Active Reflection Coefficient (TARC). The results indicate that including a ring resonator in the antenna design significantly improves the antenna's performance, and machine learning algorithms, particularly Random Forest, can effectively predict and optimize critical parameters for antenna design in 5G applications.
{"title":"Machine Learning Assisted mm-Wave MIMO Antenna Design with High Isolation for 5G Applications","authors":"Ramasamy R;Rajavel V;Rachit Jain","doi":"10.1109/TLA.2025.11231218","DOIUrl":"https://doi.org/10.1109/TLA.2025.11231218","url":null,"abstract":"This study investigates the design and performance of millimeter-wave (mm-Wave) Multiple-Input Multiple-Output (MIMO) antennas for fifth-generation (5G) applications, with a particular focus on the consequences of incorporating a ring resonator within the antenna system. This study compares two design variations, one with a ring resonator, and one without to assess their impact on enhancing the antenna's performance characteristics. The research employs five machine learning algorithms, namely, Decision Tree, Random Forest, K-Nearest Neighbors (KNN), XG-Boost, and Gradient Boosting Regressor (GBR), to estimate return loss. Among these, the Random Forest algorithm demonstrates superior performance in terms of accuracy, Mean Squared Error (MSE), Mean Absolute Percentage Error (MAPE), Mean Absolute Error (MAE), and R-squared metrics. The proposed MIMO antenna system shows better performance in Envelope Correlation Coefficient (ECC), Diversity Gain (DG), Channel Capacity Loss (CCL) and Total Active Reflection Coefficient (TARC). The results indicate that including a ring resonator in the antenna design significantly improves the antenna's performance, and machine learning algorithms, particularly Random Forest, can effectively predict and optimize critical parameters for antenna design in 5G applications.","PeriodicalId":55024,"journal":{"name":"IEEE Latin America Transactions","volume":"23 12","pages":"1325-1334"},"PeriodicalIF":1.3,"publicationDate":"2025-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11231218","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145455889","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-07DOI: 10.1109/TLA.2025.11231223
Sílvia Costa Ferreira;Ronnielli Chagas de Oliveira;Alexandre de Araújo;Alexandre Luiz da Silva;Marcelo Arriel Rezende;João Paulo de Carvalho Pedroso;Joaquim Paulo da Silva
The increasing integration of Distributed Energy Resources (DERs) in power distribution networks demands accurate system modeling and reliable power flow analysis. This paper presents a structured methodology for modeling a real Electrical Distribution System (EDS) using OpenDSS, applied to the Federal University of Lavras (UFLA), Brazil. Due to the lack of georeferenced data, the method combines satellite-based geolocation with load characterization from real measurements and statistical distributions. The model supports deterministic and time-series power flow simulations under three conditions: without DERs, with a 1.2 MWp photovoltaic plant, and with additional power factor correction. To achieve this, an incremental algorithm is proposed to determine the optimal size of a fixed capacitor bank at the feeder, improving the power factor without violating constraints. Results showed that DER integration reverses power flow, increases losses, and reduces the power factor, which also becomes variable and highly dependent on photovoltaic generation, but is improved by the proposed algorithm. This methodology enables effective, simplified modeling and analysis of real-world EDSs.
{"title":"Modeling and Analysis of Distribution Power System at UFLA Using OpenDSS","authors":"Sílvia Costa Ferreira;Ronnielli Chagas de Oliveira;Alexandre de Araújo;Alexandre Luiz da Silva;Marcelo Arriel Rezende;João Paulo de Carvalho Pedroso;Joaquim Paulo da Silva","doi":"10.1109/TLA.2025.11231223","DOIUrl":"https://doi.org/10.1109/TLA.2025.11231223","url":null,"abstract":"The increasing integration of Distributed Energy Resources (DERs) in power distribution networks demands accurate system modeling and reliable power flow analysis. This paper presents a structured methodology for modeling a real Electrical Distribution System (EDS) using OpenDSS, applied to the Federal University of Lavras (UFLA), Brazil. Due to the lack of georeferenced data, the method combines satellite-based geolocation with load characterization from real measurements and statistical distributions. The model supports deterministic and time-series power flow simulations under three conditions: without DERs, with a 1.2 MWp photovoltaic plant, and with additional power factor correction. To achieve this, an incremental algorithm is proposed to determine the optimal size of a fixed capacitor bank at the feeder, improving the power factor without violating constraints. Results showed that DER integration reverses power flow, increases losses, and reduces the power factor, which also becomes variable and highly dependent on photovoltaic generation, but is improved by the proposed algorithm. This methodology enables effective, simplified modeling and analysis of real-world EDSs.","PeriodicalId":55024,"journal":{"name":"IEEE Latin America Transactions","volume":"23 12","pages":"1261-1270"},"PeriodicalIF":1.3,"publicationDate":"2025-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11231223","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145455984","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Automatic Modulation Recognition (AMR) is a fundamental capability for Unmanned Aerial Vehicle (UAV) communication systems in sixth-generation (6G) wireless networks. It enables UAVs to intelligently identify and track received signals, supporting reliable connectivity under dynamic environments. In practical UAV applications, AMR methods must achieve high recognition accuracy with minimal computational complexity, since UAV platforms operate under strict constraints in storage, memory, and processing power. While recent Deep Learning (DL)-based solutions have advanced AMR performance, most prioritize accuracy at the cost of significantly larger models and higher computational demands. Conversely, lightweight models often lack the accuracy required for real-time deployment, limiting their practical utility. To overcome these limitations, this paper presents a novel Super Light Convolutional Neural Network (SLCNN) for AMR. Unlike conventional models, SLCNN em-ploys a carefully optimized architecture with fewer convolutional layers, smaller filters, and pooling operations, combined with Gaussian noise and dropout for robust generalization. This design strategy reduces model size and inference time while preserving high accuracy. The proposed SLCNN was evaluated on the HisarMod 2019.1 dataset and validated across RML 2016.10a, 2016.10b, and 2018.01a datasets. Experimental comparisons with Convolutional Long Short-Term Memory Deep Neural Network (CLNN), Long Short-Term Memory, Gated Recurrent Unit, and Residual Network highlight that SLCNN achieves superior results, attaining 98.50% classification accuracy with significantly reduced computational cost. Furthermore, deployment on the NVIDIA Jetson Orin Nano demonstrates real-time suitability, confirming the models effectiveness for UAV-based 6G wireless networks.
自动调制识别（AMR）是第六代（6G）无线网络中无人机（UAV）通信系统的一项基本能力。它使无人机能够智能地识别和跟踪接收到的信号，支持动态环境下的可靠连接。在实际的无人机应用中，AMR方法必须以最小的计算复杂度实现高识别精度，因为无人机平台在存储、内存和处理能力方面受到严格限制。虽然最近基于深度学习（DL）的解决方案具有先进的AMR性能，但大多数解决方案以更大的模型和更高的计算需求为代价来优先考虑准确性。相反，轻量级模型通常缺乏实时部署所需的准确性，限制了它们的实际用途。为了克服这些限制，本文提出了一种用于AMR的超轻卷积神经网络（SLCNN）。与传统模型不同，SLCNN采用了精心优化的架构，具有更少的卷积层，更小的滤波器和池化操作，并结合高斯噪声和dropout进行鲁棒泛化。这种设计策略减少了模型大小和推理时间，同时保持了较高的准确性。在HisarMod 2019.1数据集上对所提出的SLCNN进行了评估，并在RML 2016.10a、2016.10b和2018.01a数据集上进行了验证。与卷积长短期记忆深度神经网络（Convolutional Long - Short-Term Memory Deep Neural Network， CLNN）、长短期记忆、门控递归单元（Gated Recurrent Unit）和残差网络（Residual Network）的实验比较表明，SLCNN在分类准确率达到98.50%的同时显著降低了计算成本。此外，在NVIDIA Jetson Orin Nano上的部署验证了该模型的实时性，证实了该模型在基于无人机的6G无线网络中的有效性。
{"title":"A Super Light Convolutional Neural Network for Automatic Modulation Recognition in Unmanned Aerial Vehicles based 6G Wireless Network","authors":"Debbarni Sarkar;Samarth Verma;Rupa Kumari;Yogita Yogita;Vipin Pal;Satyendra Singh Yadav","doi":"10.1109/TLA.2025.11231226","DOIUrl":"https://doi.org/10.1109/TLA.2025.11231226","url":null,"abstract":"Automatic Modulation Recognition (AMR) is a fundamental capability for Unmanned Aerial Vehicle (UAV) communication systems in sixth-generation (6G) wireless networks. It enables UAVs to intelligently identify and track received signals, supporting reliable connectivity under dynamic environments. In practical UAV applications, AMR methods must achieve high recognition accuracy with minimal computational complexity, since UAV platforms operate under strict constraints in storage, memory, and processing power. While recent Deep Learning (DL)-based solutions have advanced AMR performance, most prioritize accuracy at the cost of significantly larger models and higher computational demands. Conversely, lightweight models often lack the accuracy required for real-time deployment, limiting their practical utility. To overcome these limitations, this paper presents a novel Super Light Convolutional Neural Network (SLCNN) for AMR. Unlike conventional models, SLCNN em-ploys a carefully optimized architecture with fewer convolutional layers, smaller filters, and pooling operations, combined with Gaussian noise and dropout for robust generalization. This design strategy reduces model size and inference time while preserving high accuracy. The proposed SLCNN was evaluated on the HisarMod 2019.1 dataset and validated across RML 2016.10a, 2016.10b, and 2018.01a datasets. Experimental comparisons with Convolutional Long Short-Term Memory Deep Neural Network (CLNN), Long Short-Term Memory, Gated Recurrent Unit, and Residual Network highlight that SLCNN achieves superior results, attaining 98.50% classification accuracy with significantly reduced computational cost. Furthermore, deployment on the NVIDIA Jetson Orin Nano demonstrates real-time suitability, confirming the models effectiveness for UAV-based 6G wireless networks.","PeriodicalId":55024,"journal":{"name":"IEEE Latin America Transactions","volume":"23 12","pages":"1305-1317"},"PeriodicalIF":1.3,"publicationDate":"2025-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11231226","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145455913","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-07DOI: 10.1109/TLA.2025.11231229
Rayappa David Amar Raj;Rama Muni Reddy Yanamala;Archana Pallakonda;Anil Naik Kanasottu
Solar PV arrays are susceptible to various faults, such as hotspots, cracks, and Potential Induced Degradation, which can impair efficiency and longevity. Traditional fault detection methods are time-intensive and limited in accuracy, especially for large-scale installations. This paper proposes an Adaptive Attention Pyramid Network (AAPN) for accurate and efficient fault detection in PV modules. AAPN integrates depthwise separable convolutions, squeeze-and-excitation blocks, and adaptive attention mechanisms to achieve high accuracy in identifying fault types across different classification complexities. Extensive experimentation on a comprehensive dataset of infrared PV images, organized into 12 fault classes, demonstrated AAPNs high classification accuracy of up to 96% in binary and 92% in 12-class classification scenarios. The proposed model is tested using an infrared solar module dataset for 2-class, 8- class, 11-class, and 12-class fault categories. Its effectiveness is compared with 69 existing deep-learning models for various fault classes. An ablation study was conducted to evaluate the impact of different architectural components, such as depthwise separable convolutions and squeeze-and-excitation blocks, on the models performance, showing an optimal trade-off between accuracy and computational efficiency. The proposed architecture model is very lightweight, utilizing only 0.8 million parameters. Its effective balance between high accuracies and low parameter utilization makes it highly suitable for deployment on drone-based edge devices, facilitating on-site real-time PV fault monitoring, maintenance, and detection. Additionally, the model has been successfully implemented on the Google Coral Edge TPU, achieving 40.2 ms inference time per image, confirming its efficiency and suitability for real-time applications in resource-constrained environments.
{"title":"AAPN-Tiny: A Compact Edge-Deployable Adaptive Attention Pyramid Architecture for Multi-Class Fault Diagnosis in Solar Photovoltaic Modules","authors":"Rayappa David Amar Raj;Rama Muni Reddy Yanamala;Archana Pallakonda;Anil Naik Kanasottu","doi":"10.1109/TLA.2025.11231229","DOIUrl":"https://doi.org/10.1109/TLA.2025.11231229","url":null,"abstract":"Solar PV arrays are susceptible to various faults, such as hotspots, cracks, and Potential Induced Degradation, which can impair efficiency and longevity. Traditional fault detection methods are time-intensive and limited in accuracy, especially for large-scale installations. This paper proposes an Adaptive Attention Pyramid Network (AAPN) for accurate and efficient fault detection in PV modules. AAPN integrates depthwise separable convolutions, squeeze-and-excitation blocks, and adaptive attention mechanisms to achieve high accuracy in identifying fault types across different classification complexities. Extensive experimentation on a comprehensive dataset of infrared PV images, organized into 12 fault classes, demonstrated AAPNs high classification accuracy of up to 96% in binary and 92% in 12-class classification scenarios. The proposed model is tested using an infrared solar module dataset for 2-class, 8- class, 11-class, and 12-class fault categories. Its effectiveness is compared with 69 existing deep-learning models for various fault classes. An ablation study was conducted to evaluate the impact of different architectural components, such as depthwise separable convolutions and squeeze-and-excitation blocks, on the models performance, showing an optimal trade-off between accuracy and computational efficiency. The proposed architecture model is very lightweight, utilizing only 0.8 million parameters. Its effective balance between high accuracies and low parameter utilization makes it highly suitable for deployment on drone-based edge devices, facilitating on-site real-time PV fault monitoring, maintenance, and detection. Additionally, the model has been successfully implemented on the Google Coral Edge TPU, achieving 40.2 ms inference time per image, confirming its efficiency and suitability for real-time applications in resource-constrained environments.","PeriodicalId":55024,"journal":{"name":"IEEE Latin America Transactions","volume":"23 12","pages":"1284-1296"},"PeriodicalIF":1.3,"publicationDate":"2025-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11231229","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145455849","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-07DOI: 10.1109/TLA.2025.11231225
Dalvana Lopes Ribeiro;André Andrade Longaray
This article describes the investigation of scientific production on the combination of geographic information systems (GIS) and multicriteria methods (MCDM/A). For this, the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) protocol was followed. As a result, from 8.001 initial records, a portfolio of 97 articles aligned with the research theme was selected, using the Web of Science and Scopus databases. The analysis was conducted with the support of the CiteSpace, VOSviewer and Bibliometrix R package software, allowing us to understand the trends and development of the areas over time. The results point to significant challenges, such as the need for more intuitive interfaces, training of management teams, greater flexibility of systems and difficulties in operationalizing decision-making. This study contributes to the understanding of the evolution of the field and highlights gaps that can guide future research.
本文介绍了地理信息系统（GIS）与多标准方法（MCDM/A）相结合的科学生产研究。为此，遵循系统评价和荟萃分析首选报告项目（PRISMA）协议。结果，从8.001条初始记录中，使用Web of Science和Scopus数据库，选择了97篇符合研究主题的文章组合。该分析是在CiteSpace、VOSviewer和Bibliometrix R软件包软件的支持下进行的，使我们能够了解这些领域随着时间的推移的趋势和发展。结果指出了重大的挑战，例如需要更直观的界面、训练管理队、系统的更大灵活性和决策的操作困难。这项研究有助于理解该领域的演变，并突出了可以指导未来研究的空白。
{"title":"The use of geographic information systems combined with multicriteria methods in organizations: a systematic literature review","authors":"Dalvana Lopes Ribeiro;André Andrade Longaray","doi":"10.1109/TLA.2025.11231225","DOIUrl":"https://doi.org/10.1109/TLA.2025.11231225","url":null,"abstract":"This article describes the investigation of scientific production on the combination of geographic information systems (GIS) and multicriteria methods (MCDM/A). For this, the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) protocol was followed. As a result, from 8.001 initial records, a portfolio of 97 articles aligned with the research theme was selected, using the Web of Science and Scopus databases. The analysis was conducted with the support of the CiteSpace, VOSviewer and Bibliometrix R package software, allowing us to understand the trends and development of the areas over time. The results point to significant challenges, such as the need for more intuitive interfaces, training of management teams, greater flexibility of systems and difficulties in operationalizing decision-making. This study contributes to the understanding of the evolution of the field and highlights gaps that can guide future research.","PeriodicalId":55024,"journal":{"name":"IEEE Latin America Transactions","volume":"23 12","pages":"1172-1188"},"PeriodicalIF":1.3,"publicationDate":"2025-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11231225","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145456001","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-07DOI: 10.1109/TLA.2025.11231215
Gabriel Rodrigues Santos;Eduardo Zancul;Erik Eduardo Rego
Electrical substations must keep their critical task of providing the power grid safe, reliable, protected, and manageable electricity flow within the context of increased digitalization in the power sector. On one hand, digital substation automation systems enable novel capabilities and functions, but on the other hand utilities must effectively manage such transformation while keeping their assets in operation. This paper presents an overview of how substations of different epochs are designed, operated and maintained in a practice-centered context of a Brazilian transmission utility. Drawing from a comparative case study based on a theoretical classification model, three high-voltage substations with different degrees of digitalization are analyzed regarding their automation system's design, features, lifetime upgrades, and future implications. The study shows that real-world conditions present challenges to operators and utilities retrofit existing substations in modernization efforts. There is a strong tendency to digitalize substation automation systems based on the IEC 61850, but the implementation of a process bus is still not widespread. Contrasting the cases with the academic literature reveals there are still areas that require further development to be competitively implemented by utilities, such as the usage of low-power instrument transformers, and that utilities must actively prepare to leverage the long-term benefits of those installations. Particularly in Brazil, extensive upcoming investment in such facilities are expected. As such, this study contributes to the understanding and discussion of the role of digital technologies in substations and their relationship to the professional practice of transmission utilities.
{"title":"Pathways to digital substations: a comparative case study","authors":"Gabriel Rodrigues Santos;Eduardo Zancul;Erik Eduardo Rego","doi":"10.1109/TLA.2025.11231215","DOIUrl":"https://doi.org/10.1109/TLA.2025.11231215","url":null,"abstract":"Electrical substations must keep their critical task of providing the power grid safe, reliable, protected, and manageable electricity flow within the context of increased digitalization in the power sector. On one hand, digital substation automation systems enable novel capabilities and functions, but on the other hand utilities must effectively manage such transformation while keeping their assets in operation. This paper presents an overview of how substations of different epochs are designed, operated and maintained in a practice-centered context of a Brazilian transmission utility. Drawing from a comparative case study based on a theoretical classification model, three high-voltage substations with different degrees of digitalization are analyzed regarding their automation system's design, features, lifetime upgrades, and future implications. The study shows that real-world conditions present challenges to operators and utilities retrofit existing substations in modernization efforts. There is a strong tendency to digitalize substation automation systems based on the IEC 61850, but the implementation of a process bus is still not widespread. Contrasting the cases with the academic literature reveals there are still areas that require further development to be competitively implemented by utilities, such as the usage of low-power instrument transformers, and that utilities must actively prepare to leverage the long-term benefits of those installations. Particularly in Brazil, extensive upcoming investment in such facilities are expected. As such, this study contributes to the understanding and discussion of the role of digital technologies in substations and their relationship to the professional practice of transmission utilities.","PeriodicalId":55024,"journal":{"name":"IEEE Latin America Transactions","volume":"23 12","pages":"1297-1304"},"PeriodicalIF":1.3,"publicationDate":"2025-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11231215","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145455809","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-07DOI: 10.1109/TLA.2025.11231214
Xuan Nie;Teng Li;Yinan Yuan;Zichen Yan;Yiwen Liu;Guangpu Zhou;Bosong Chai
Coronary heart disease is one of the most common cardiovascular diseases. Currently, CTA imaging has become the most widely used modality for its diagnosis. The detection of coronary plaque is an important basis for accurate diagnosis. In order to further improve the accuracy and efficiency of coronary plaque detection, this study proposes a series of automatic detection methods for coronary plaque based on deep learning. The approach begins by segmenting coronary arteries using a Transformer model integrated with an multi-resolution overlapping attention mechanism, thereby reducing interference in plaque detection. Subsequently, a two-stage hybrid strategy is employed for centerline extraction and optimization, and a multi-angle straightened surface reconstruction method is proposed to generate high-quality data for plaque detection. An improved RetinaNet (CoroJARetinaNet) is developed, integrating an attention mechanism, enhanced feature pyramid networks, and optimized post-processing strategies. Experimental results demonstrate that the proposed method significantly improves the accuracy and efficiency of coronary plaque detection compared to traditional approaches.
{"title":"CoroJARetinaNet: A Multiscale Attention-Guided Framework for Automated Coronary Plaque Detection in CTA Images","authors":"Xuan Nie;Teng Li;Yinan Yuan;Zichen Yan;Yiwen Liu;Guangpu Zhou;Bosong Chai","doi":"10.1109/TLA.2025.11231214","DOIUrl":"https://doi.org/10.1109/TLA.2025.11231214","url":null,"abstract":"Coronary heart disease is one of the most common cardiovascular diseases. Currently, CTA imaging has become the most widely used modality for its diagnosis. The detection of coronary plaque is an important basis for accurate diagnosis. In order to further improve the accuracy and efficiency of coronary plaque detection, this study proposes a series of automatic detection methods for coronary plaque based on deep learning. The approach begins by segmenting coronary arteries using a Transformer model integrated with an multi-resolution overlapping attention mechanism, thereby reducing interference in plaque detection. Subsequently, a two-stage hybrid strategy is employed for centerline extraction and optimization, and a multi-angle straightened surface reconstruction method is proposed to generate high-quality data for plaque detection. An improved RetinaNet (CoroJARetinaNet) is developed, integrating an attention mechanism, enhanced feature pyramid networks, and optimized post-processing strategies. Experimental results demonstrate that the proposed method significantly improves the accuracy and efficiency of coronary plaque detection compared to traditional approaches.","PeriodicalId":55024,"journal":{"name":"IEEE Latin America Transactions","volume":"23 12","pages":"1152-1162"},"PeriodicalIF":1.3,"publicationDate":"2025-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11231214","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145455810","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-07DOI: 10.1109/TLA.2025.11231193
José F. Flores;H´éctor A. Moreno;Isela G. Carrera;Jose Luis Ordoñez-Avila
This paper presents the design of a novel leg-wheel transformable mechanism. The purpose of this mechanism is to combine the efficiency of wheeled locomotion with the ability of the legs to traverse difficult terrain. The mechanism has two degrees of freedom with decoupled kinematics. The decoupled kinematics of the mechanism allows the rotation and extension/flexion motions to be controlled independently by a pair of actuators, which provides some advantages. In this work, the direct kinematics of the mechanism is solved analytically. On the other hand, due to the complexity of the inverse kinematics, two different numerical methods were evaluated for solving this problem. A model based on neural networks was successfully implemented in a physical prototype to generate the trajectory of a gait cycle in the leg mode of the mechanism.
{"title":"Design and Computational Modeling of a Leg-Wheel Transformable Mechanism with Decoupled Kinematics","authors":"José F. Flores;H´éctor A. Moreno;Isela G. Carrera;Jose Luis Ordoñez-Avila","doi":"10.1109/TLA.2025.11231193","DOIUrl":"https://doi.org/10.1109/TLA.2025.11231193","url":null,"abstract":"This paper presents the design of a novel leg-wheel transformable mechanism. The purpose of this mechanism is to combine the efficiency of wheeled locomotion with the ability of the legs to traverse difficult terrain. The mechanism has two degrees of freedom with decoupled kinematics. The decoupled kinematics of the mechanism allows the rotation and extension/flexion motions to be controlled independently by a pair of actuators, which provides some advantages. In this work, the direct kinematics of the mechanism is solved analytically. On the other hand, due to the complexity of the inverse kinematics, two different numerical methods were evaluated for solving this problem. A model based on neural networks was successfully implemented in a physical prototype to generate the trajectory of a gait cycle in the leg mode of the mechanism.","PeriodicalId":55024,"journal":{"name":"IEEE Latin America Transactions","volume":"23 12","pages":"1219-1229"},"PeriodicalIF":1.3,"publicationDate":"2025-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11231193","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145455888","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-07DOI: 10.1109/TLA.2025.11231222
Vinícius Di Oliveira;Pedro Carvalho Brom;Li Weigang
This study addresses a limitation in Retrieval-Augmented Generation (RAG) systems: poor retrieval accuracy when vague prompts or metadata are missing. We propose the Two-Step RAG method to overcome this. The first step performs a broad semantic search. The second uses an LLM to extract structured metadata to refine results through contextual filtering. This structure balances coverage and precision, proving effective in well-structured domains such as the Mercosur Common Nomenclature (NCM). The method is evaluated using a bootstrap-based multivariate linear mixed model, accounting for variability in temperature, top-p and prompt formulation. Two-Step RAG improves quality by a factor of 1.94, agreement by 2.31 and accuracy by 2.51 on average, while reducing hallucination to 0.82x compared to conventional RAG. It also shows reduced output variability in high-performing models, with coefficients of variation in quality dropping to 3033% for gpt-4o-mini and deepseek-chat. These models achieve the best results, with accuracy gains exceeding 3x and hallucination reduced to 3655% of the baseline. The method is robust across configurations and offers practical value for applications requiring high retrieval precision.
{"title":"Two-Step RAG for Metadata Filtering and Statistical LLM Evaluation","authors":"Vinícius Di Oliveira;Pedro Carvalho Brom;Li Weigang","doi":"10.1109/TLA.2025.11231222","DOIUrl":"https://doi.org/10.1109/TLA.2025.11231222","url":null,"abstract":"This study addresses a limitation in Retrieval-Augmented Generation (RAG) systems: poor retrieval accuracy when vague prompts or metadata are missing. We propose the Two-Step RAG method to overcome this. The first step performs a broad semantic search. The second uses an LLM to extract structured metadata to refine results through contextual filtering. This structure balances coverage and precision, proving effective in well-structured domains such as the Mercosur Common Nomenclature (NCM). The method is evaluated using a bootstrap-based multivariate linear mixed model, accounting for variability in temperature, top-p and prompt formulation. Two-Step RAG improves quality by a factor of 1.94, agreement by 2.31 and accuracy by 2.51 on average, while reducing hallucination to 0.82x compared to conventional RAG. It also shows reduced output variability in high-performing models, with coefficients of variation in quality dropping to 3033% for gpt-4o-mini and deepseek-chat. These models achieve the best results, with accuracy gains exceeding 3x and hallucination reduced to 3655% of the baseline. The method is robust across configurations and offers practical value for applications requiring high retrieval precision.","PeriodicalId":55024,"journal":{"name":"IEEE Latin America Transactions","volume":"23 12","pages":"1201-1210"},"PeriodicalIF":1.3,"publicationDate":"2025-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11231222","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145455911","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：