期刊文献+
共找到4篇文章
< 1 >
每页显示 20 50 100
Process Monitoring and Terminal Verification of Cable-Stayed Bridges with Corrugated Steel Webs under Contruction
1
作者 Kexin Zhang xinyuan shen +1 位作者 Longsheng Bao He Liu 《Structural Durability & Health Monitoring》 EI 2023年第2期131-158,共28页
In this paper,the construction process of a cable-stayed bridge with corrugated steel webs was monitored.Moreover,the end performance of the bridge was verified by load test.Owing to the consideration of the bridge st... In this paper,the construction process of a cable-stayed bridge with corrugated steel webs was monitored.Moreover,the end performance of the bridge was verified by load test.Owing to the consideration of the bridge structure safety,it is necessary to monitor the main girder deflection,stress,construction error and safety state during construction.Furthermore,to verify whether the bridge can meet the design requirements,the static and dynamic load tests are carried out after the completion of the bridge.The results of construction monitoring show that the stress state of the structure during construction is basically consistent with the theoretical calculation and design requirements,and both meet the design and specification requirements.The final measured stress state of the structure is within the allowable range of the cable-stayed bridge,and the stress state of the structure is normal and meets the specification requirements.The results of load tests show that the measured deflection values of the mid-span section of the main girder are less than the theoretical calculation values.The maximum deflection of the girder is−20.90 mm,which is less than−22.00 mm of the theoretical value,indicating that the girder has sufficient structural stiffness.The maximum impact coefficient under dynamic load test is 1.08,which is greater than 1.05 of theoretical value,indicating that the impact effect of heavy-duty truck on this type of bridge is larger.This study can provide important reference value for construction and maintenance of similar corrugated steel web cable-stayed bridges. 展开更多
关键词 Cable-stayed bridge corrugated steel web construction monitoring static load test dynamic load test
下载PDF
Recent advances of bioresponsive polymeric nanomedicine for cancer therapy
2
作者 Tu Hong xinyuan shen +8 位作者 Madiha Zahra Syeda Yang Zhang Haonan sheng Yipeng Zhou JinMing Xu Chaojie Zhu Hongjun Li Zhen Gu Longguang Tang 《Nano Research》 SCIE EI CSCD 2023年第2期2660-2671,共12页
A bioresponsive polymeric nanocarrier for drug delivery is able to alter its physical and physicochemical properties in response to a variety of biological signals and pathological changes,and can exert its therapeuti... A bioresponsive polymeric nanocarrier for drug delivery is able to alter its physical and physicochemical properties in response to a variety of biological signals and pathological changes,and can exert its therapeutic efficacy within a confined space.These nanosystems can optimize the biodistribution and subcellular location of therapeutics by exploiting the differences in biochemical properties between tumors and normal tissues.Moreover,bioresponsive polymer-based nanosystems could be rationally designed as precision therapeutic platforms by optimizing the combination of responsive elements and therapeutic components according to the patient-specific disease type and stage.In this review,recent advances in smart bioresponsive polymeric nanosystems for cancer chemotherapy and immunotherapy will be summarized.We mainly discuss three categories,including acidity-sensitive,redox-responsive,and enzyme-triggered polymeric nanosystems.The important issues regarding clinical translation such as reproducibility,manufacture,and probable toxicity,are also commented. 展开更多
关键词 drug delivery POLYMER bioresponsive IMMUNOTHERAPY cancer therapy
原文传递
面向异构数据的自适应个性化联邦学习——一种基于参数分解和持续学习的方法 被引量:2
3
作者 倪宣明 沈鑫圆 张海 《中国科学:信息科学》 CSCD 北大核心 2022年第12期2306-2320,共15页
联邦学习允许资源受限的边缘计算设备协作训练机器学习模型,同时能够保证数据不离开本地设备,但也面临着异构数据下全局模型收敛缓慢甚至偏离最优解的挑战.为解决上述问题,本文提出一种自适应个性化联邦学习(adaptive personalized fede... 联邦学习允许资源受限的边缘计算设备协作训练机器学习模型,同时能够保证数据不离开本地设备,但也面临着异构数据下全局模型收敛缓慢甚至偏离最优解的挑战.为解决上述问题,本文提出一种自适应个性化联邦学习(adaptive personalized federated learning,APFL)算法,在同时包括空间和时间维度的多任务学习框架下,考虑面向异构数据的联邦优化问题.首先,APFL采用参数分解策略,将待训练模型参数分解为全局共享参数和客户端特定参数,在提取所有客户端公共知识的同时实现针对每个客户端的个性化建模.进一步地,APFL将每个客户端上执行的局部优化构建为顺序多任务学习,通过对全局共享参数的更新施加弹性权重巩固(elastic weight consolidation,EWC)惩罚,实现了全局共享模型中重要参数的记忆保留和非重要参数的快速学习.多个联邦基准数据集上的对比实验验证了本文方法的有效性和优越性. 展开更多
关键词 联邦学习 边缘计算 异构数据 多任务学习 持续学习 参数分解 个性化
原文传递
基于联合分布核适配的迁移学习及其隐私保护
4
作者 倪宣明 沈鑫圆 张海 《中国科学:信息科学》 CSCD 北大核心 2021年第10期1609-1624,共16页
迁移学习利用不同但相关的源域标记数据来解决目标领域的学习问题,大多数减小域间分布差异的方法依赖于最大均值差异距离,但其仅仅能匹配域间数据分布的各阶矩.此外,隐私保护意识的增强限制了对数据源的访问,对迁移学习的发展提出了新... 迁移学习利用不同但相关的源域标记数据来解决目标领域的学习问题,大多数减小域间分布差异的方法依赖于最大均值差异距离,但其仅仅能匹配域间数据分布的各阶矩.此外,隐私保护意识的增强限制了对数据源的访问,对迁移学习的发展提出了新的挑战.本文提出一种基于联合分布核适配的迁移学习及其隐私保护方法,直接在再生核希尔伯特空间中同时减小域间边缘分布和条件分布的差异,从而学习一个域不变核矩阵.此外,我们设置数据源双方首先访问一个相同的随机投影函数,然后聚合器发布基于目标扰动的差分隐私核分类器,在实现基于核的联合分布适配的同时,避免了数据源与聚合器直接共享原始特征数据.在多个文本和图像迁移学习基准数据集上进行了对比实验和参数分析,结果显示本文方法具有良好的有效性. 展开更多
关键词 迁移学习 隐私保护 分布适配 谱学习 差分隐私
原文传递
上一页 1 下一页 到第
使用帮助 返回顶部