期刊文献+
共找到2篇文章
< 1 >
每页显示 20 50 100
Topology Optimization for Harmonic Excitation Structures with Minimum Length Scale Control Using the Discrete Variable Method
1
作者 Hongliang Liu peijin wang +2 位作者 Yuan Liang Kai Long Dixiong Yang 《Computer Modeling in Engineering & Sciences》 SCIE EI 2023年第6期1941-1964,共24页
Continuumtopology optimization considering the vibration response is of great value in the engineering structure design.The aimof this studyis toaddress the topological designoptimizationof harmonic excitationstructur... Continuumtopology optimization considering the vibration response is of great value in the engineering structure design.The aimof this studyis toaddress the topological designoptimizationof harmonic excitationstructureswith minimumlength scale control to facilitate structuralmanufacturing.Astructural topology design based on discrete variables is proposed to avoid localized vibration modes,gray regions and fuzzy boundaries in harmonic excitation topology optimization.The topological design model and sensitivity formulation are derived.The requirement of minimum size control is transformed into a geometric constraint using the discrete variables.Consequently,thin bars,small holes,and sharp corners,which are not conducive to the manufacturing process,can be eliminated from the design results.The present optimization design can efficiently achieve a 0–1 topology configuration with a significantly improved resonance frequency in a wide range of excitation frequencies.Additionally,the optimal solution for harmonic excitation topology optimization is not necessarily symmetric when the load and support are symmetric,which is a distinct difference fromthe static optimization design.Hence,one-half of the design domain cannot be selected according to the load and support symmetry.Numerical examples are presented to demonstrate the effectiveness of the discrete variable design for excitation frequency topology optimization,and to improve the design manufacturability. 展开更多
关键词 Discrete variable topology optimization harmonic excitation minimumlength scale control geometric constraint MANUFACTURABILITY
下载PDF
遥感跨模态智能解译:模型、数据与应用 被引量:3
2
作者 付琨 王佩瑾 +5 位作者 冯瑛超 李俊希 何琪彬 肖思宁 刁文辉 孙显 《中国科学:信息科学》 CSCD 北大核心 2023年第8期1529-1559,共31页
以深度学习为代表的人工智能技术已被广泛应用于遥感图像解译中.相比自然场景图像,遥感图像具有载荷类型多、成像机理差异大等特点,使得现有面向单传感器、纯数据驱动的智能解译方法应用到不同模态数据时,性能上限难以突破.尤其在面向... 以深度学习为代表的人工智能技术已被广泛应用于遥感图像解译中.相比自然场景图像,遥感图像具有载荷类型多、成像机理差异大等特点,使得现有面向单传感器、纯数据驱动的智能解译方法应用到不同模态数据时,性能上限难以突破.尤其在面向多传感获取的、大范围的、目标种类较多的复杂应用场景时,实际性能受限更为严重.本文主要对遥感智能解译结合多模态数据和多任务学习的研究工作进行综述,重点从基本概念、研究方法和应用场景3个方面进行展开.并且介绍了基于分域提取和跨域融合理念设计的模型架构,通过从海量多模态数据中提取通用特征,实现单个基础模型完成多类下游任务的泛化解译,在不同模态解译任务中表现优异,并实际应用推广.最后,对遥感多模态多任务学习未来技术发展方向进行展望. 展开更多
关键词 深度学习 多模态 多任务学习 基础模型 自监督训练
原文传递
上一页 1 下一页 到第
使用帮助 返回顶部