随着变电站智能化运维需求的不断增长,开关状态的精准识别变得尤为重要。为了解决变电站开关状态识别准确性不足和效率低下的问题,本研究基于You Only Look Once version 5s(YOLOv5s)模型,开展了智能运维技术的优化研究。研究首先深入...随着变电站智能化运维需求的不断增长,开关状态的精准识别变得尤为重要。为了解决变电站开关状态识别准确性不足和效率低下的问题,本研究基于You Only Look Once version 5s(YOLOv5s)模型,开展了智能运维技术的优化研究。研究首先深入分析了运维系统的需求,并收集了变电站开关图像数据,随后针对YOLOv5s模型的特征提取网络、融合网络和损失函数等进行了一系列优化。通过引入自适应空间特征融合模块、通道注意力机制以及改进的损失函数,构建了一个更加高效且精准的模型。实验结果显示,优化后的模型在推理时间明显更短,比其他模型缩短达到5%以上,能够更迅速地对开关状态做出判断。而最终检测精度达84.9%,相较于其他模型提高了1.8%以上,充分彰显了其对开关状态识别的高准确性。此外,召回率也有着近20%及以上的显著提升。实验结果表明,优化模型不仅能高效识别开关状态,对变电站智能运维也有积极推动作用。展开更多
智能视觉感知与理解技术是人工智能、智慧城市、智慧交通、智能安防等领域的重要研究内容,具有重要的战略意义。介绍了各国智能视觉感知与理解相关政策规划,统计分析了中国国家自然科学基金相关规划项目;对大范围视频视觉感知与理解领...智能视觉感知与理解技术是人工智能、智慧城市、智慧交通、智能安防等领域的重要研究内容,具有重要的战略意义。介绍了各国智能视觉感知与理解相关政策规划,统计分析了中国国家自然科学基金相关规划项目;对大范围视频视觉感知与理解领域以及无人平台视觉感知与理解领域的Web of Science数据库收录论文进行了分析,对IBM、Google、华为、百度等领域主要公司相关产品进行了调研;从学术和产业两个角度,分别总结了智能视觉感知与理解的研究热点和发展态势。展开更多
Industrial robot application(IRA)provides an opportunity for the low-carbon development of trade.This study focuses on the green revolution of manufacturing export trade,analyzes the mechanism by which IRA affects CO_...Industrial robot application(IRA)provides an opportunity for the low-carbon development of trade.This study focuses on the green revolution of manufacturing export trade,analyzes the mechanism by which IRA affects CO_(2) emissions embodied in manufacturing exports(CIE),and conducts an empirical test based on panel data from 37 countries from 2000 to 2019.This study found that first,IRA can significantly reduce CIE,but there is a U-shaped nexus between the two,which shows a rebound effect.Second,the heterogeneity test demonstrates that in com-parison to both the low-tech and high-tech sectors,IRA in the medium-tech industry can significantly reduce CIE;compared with the low-IRA sectors,the high-IRA sectors exhibit a more obvious reduction.In addition,IRA has a stronger effect on high-carbon-intensity areas.Third,the mechanism test shows that IRA mainly affects CIE through low-carbon technology and productivity effects.Moreover,environmental regulations and the manufacturing in-telligence process positively moderate the nexus between IRA and CIE.Finally,these conclusions provide possible empirical evidence for the smart evolution of the manufacturing industry and the green development of trade.展开更多
文摘随着变电站智能化运维需求的不断增长,开关状态的精准识别变得尤为重要。为了解决变电站开关状态识别准确性不足和效率低下的问题,本研究基于You Only Look Once version 5s(YOLOv5s)模型,开展了智能运维技术的优化研究。研究首先深入分析了运维系统的需求,并收集了变电站开关图像数据,随后针对YOLOv5s模型的特征提取网络、融合网络和损失函数等进行了一系列优化。通过引入自适应空间特征融合模块、通道注意力机制以及改进的损失函数,构建了一个更加高效且精准的模型。实验结果显示,优化后的模型在推理时间明显更短,比其他模型缩短达到5%以上,能够更迅速地对开关状态做出判断。而最终检测精度达84.9%,相较于其他模型提高了1.8%以上,充分彰显了其对开关状态识别的高准确性。此外,召回率也有着近20%及以上的显著提升。实验结果表明,优化模型不仅能高效识别开关状态,对变电站智能运维也有积极推动作用。
文摘智能视觉感知与理解技术是人工智能、智慧城市、智慧交通、智能安防等领域的重要研究内容,具有重要的战略意义。介绍了各国智能视觉感知与理解相关政策规划,统计分析了中国国家自然科学基金相关规划项目;对大范围视频视觉感知与理解领域以及无人平台视觉感知与理解领域的Web of Science数据库收录论文进行了分析,对IBM、Google、华为、百度等领域主要公司相关产品进行了调研;从学术和产业两个角度,分别总结了智能视觉感知与理解的研究热点和发展态势。
基金the National Social Science Foundation of China(Grant No.23FGLB024)Special Project on“Promoting High-Quality Development through the Integration of the Yangtze River Delta”of Shaoxing University(Grant No.2024CSJYB01)to provide fund for the study。
文摘Industrial robot application(IRA)provides an opportunity for the low-carbon development of trade.This study focuses on the green revolution of manufacturing export trade,analyzes the mechanism by which IRA affects CO_(2) emissions embodied in manufacturing exports(CIE),and conducts an empirical test based on panel data from 37 countries from 2000 to 2019.This study found that first,IRA can significantly reduce CIE,but there is a U-shaped nexus between the two,which shows a rebound effect.Second,the heterogeneity test demonstrates that in com-parison to both the low-tech and high-tech sectors,IRA in the medium-tech industry can significantly reduce CIE;compared with the low-IRA sectors,the high-IRA sectors exhibit a more obvious reduction.In addition,IRA has a stronger effect on high-carbon-intensity areas.Third,the mechanism test shows that IRA mainly affects CIE through low-carbon technology and productivity effects.Moreover,environmental regulations and the manufacturing in-telligence process positively moderate the nexus between IRA and CIE.Finally,these conclusions provide possible empirical evidence for the smart evolution of the manufacturing industry and the green development of trade.
