The present work aims to develop a method for reliability-based optimum design of composite structures. A procedure combining particle swarm optimization （PSO） and finite element analysis （FEA） has been proposed. ...The present work aims to develop a method for reliability-based optimum design of composite structures. A procedure combining particle swarm optimization （PSO） and finite element analysis （FEA） has been proposed. Numerical examples for the reliability design optimization （RDO） of a laminate and a composite cylindrical shell are worked out to demonstrate the effectiveness of the method. Then a design for composite pressure vessels is studied. The advantages and necessity of RDO over the conventional equi-strength design are addressed. Examples show that the proposed method has good stability and is efficient in dealing with the probabilistic optimal design of composite structures. It may serve as an effective tool to optimize other complicated structures with uncertainties.展开更多
随着“双碳”目标的提出,天然气作为火电转型过渡时期的重要能源,亟需通过绿色转型实现安全、稳定运行。碳捕集、利用和储存(carbon capture,utilization and storage,CCUS)技术的应用将为燃气电厂低碳发展提供可靠的技术支撑。文中提...随着“双碳”目标的提出,天然气作为火电转型过渡时期的重要能源,亟需通过绿色转型实现安全、稳定运行。碳捕集、利用和储存(carbon capture,utilization and storage,CCUS)技术的应用将为燃气电厂低碳发展提供可靠的技术支撑。文中提出一种发电机组(generator unit,GU)-电转气(power to gas,P2G)-CCUS系统,并构建一个考虑风电输出不确定性的数据驱动鲁棒优化(data-driven robust optimization,DDRO)模型。在此基础上,由于现有优化方法无法实现科学成本分配,考虑到主体间的合作博弈关系及系统运行的稳定性,建立了系统内部基于核仁法(nucleolus based cooperative game,NCG)的成本分配模型,以确保成本分配的科学性和合理性。结果表明:引入CCUS可以显著减少碳排放,并通过参与碳市场获得额外的利润;DDRO模型可以有效抵抗不确定风电输出的干扰,增强系统运行的安全性,降低传统优化模型的保守性;NCG模型可以实现GU、P2G和CCUS之间的合理分配,使得每个参与者都可以获得比其独立运行时更高的收益,提高参与者的合作意愿,进而增强合作的长期性与稳定性。展开更多
目的:建立基于乳腺三维超声视频的深度学习模型(deep learning model based on three-dimensional ultrasound videos,DL-3DUV),并探讨该模型辅助不同年资医师鉴别乳腺肿块良恶性的应用价值。方法:采用ResNet50模型作为骨干网,建立乳腺...目的:建立基于乳腺三维超声视频的深度学习模型(deep learning model based on three-dimensional ultrasound videos,DL-3DUV),并探讨该模型辅助不同年资医师鉴别乳腺肿块良恶性的应用价值。方法:采用ResNet50模型作为骨干网,建立乳腺肿块良恶性的分类模型(DL-3DUV),回顾性纳入400例乳腺肿块患者(共525个病灶)的自动乳腺全容积成像(automated breast volume scanner,ABVS)动态视频,按8∶2的比例随机分为训练集和测试集,比较DL-3DUV与高年资医师、低年资医师独立或联合诊断的效能。结果:独立诊断时,DL-3DUV与高年资医师比较,灵敏度(83.33%比81.77%)、准确率(82.50%vs.84.60%)和受试者操作特征曲线的曲线下面积(area under the curve,AUC)(0.83比0.85)相当(P均>0.05),仅特异度较低(81.58%比87.73%)(P<0.05);但DL-3DUV与低年资医师比较,灵敏度(83.33%比78.60%)、特异度(81.58%比57.00%)、准确率(82.50%比68.37%)和AUC(0.83比0.68)的差异均有统计学意义(P<0.05)。高年资医师联合DL-3DUV诊断的灵敏度和AUC高于高年资医师独立诊断(89.70%比81.77%;0.91比0.85,P均<0.05),而特异度和准确率的差异无统计学意义(91.23%比87.73%;89.17%比84.60%,P均>0.05);低年资医师联合DL-3DUV诊断的诊断效能高于低年资医师独立诊断,灵敏度(88.10%比78.60%)、特异度(82.47%比57.00%)、准确率(85.47%比68.37%)和AUC(0.85比0.68)的差异均有统计学意义(P均<0.05)。结论:DL-3DUV在鉴别乳腺肿块良恶性中具有重要价值,有望成为临床辅助超声医师,尤其是低年资医师的有用工具。展开更多
基金supported by National Natural Science Foundation of China (No. 10772070)National Basic Research Program of China (No. 2011CB013800)
文摘The present work aims to develop a method for reliability-based optimum design of composite structures. A procedure combining particle swarm optimization （PSO） and finite element analysis （FEA） has been proposed. Numerical examples for the reliability design optimization （RDO） of a laminate and a composite cylindrical shell are worked out to demonstrate the effectiveness of the method. Then a design for composite pressure vessels is studied. The advantages and necessity of RDO over the conventional equi-strength design are addressed. Examples show that the proposed method has good stability and is efficient in dealing with the probabilistic optimal design of composite structures. It may serve as an effective tool to optimize other complicated structures with uncertainties.
文摘随着“双碳”目标的提出,天然气作为火电转型过渡时期的重要能源,亟需通过绿色转型实现安全、稳定运行。碳捕集、利用和储存(carbon capture,utilization and storage,CCUS)技术的应用将为燃气电厂低碳发展提供可靠的技术支撑。文中提出一种发电机组(generator unit,GU)-电转气(power to gas,P2G)-CCUS系统,并构建一个考虑风电输出不确定性的数据驱动鲁棒优化(data-driven robust optimization,DDRO)模型。在此基础上,由于现有优化方法无法实现科学成本分配,考虑到主体间的合作博弈关系及系统运行的稳定性,建立了系统内部基于核仁法(nucleolus based cooperative game,NCG)的成本分配模型,以确保成本分配的科学性和合理性。结果表明:引入CCUS可以显著减少碳排放,并通过参与碳市场获得额外的利润;DDRO模型可以有效抵抗不确定风电输出的干扰,增强系统运行的安全性,降低传统优化模型的保守性;NCG模型可以实现GU、P2G和CCUS之间的合理分配,使得每个参与者都可以获得比其独立运行时更高的收益,提高参与者的合作意愿,进而增强合作的长期性与稳定性。
文摘目的:建立基于乳腺三维超声视频的深度学习模型(deep learning model based on three-dimensional ultrasound videos,DL-3DUV),并探讨该模型辅助不同年资医师鉴别乳腺肿块良恶性的应用价值。方法:采用ResNet50模型作为骨干网,建立乳腺肿块良恶性的分类模型(DL-3DUV),回顾性纳入400例乳腺肿块患者(共525个病灶)的自动乳腺全容积成像(automated breast volume scanner,ABVS)动态视频,按8∶2的比例随机分为训练集和测试集,比较DL-3DUV与高年资医师、低年资医师独立或联合诊断的效能。结果:独立诊断时,DL-3DUV与高年资医师比较,灵敏度(83.33%比81.77%)、准确率(82.50%vs.84.60%)和受试者操作特征曲线的曲线下面积(area under the curve,AUC)(0.83比0.85)相当(P均>0.05),仅特异度较低(81.58%比87.73%)(P<0.05);但DL-3DUV与低年资医师比较,灵敏度(83.33%比78.60%)、特异度(81.58%比57.00%)、准确率(82.50%比68.37%)和AUC(0.83比0.68)的差异均有统计学意义(P<0.05)。高年资医师联合DL-3DUV诊断的灵敏度和AUC高于高年资医师独立诊断(89.70%比81.77%;0.91比0.85,P均<0.05),而特异度和准确率的差异无统计学意义(91.23%比87.73%;89.17%比84.60%,P均>0.05);低年资医师联合DL-3DUV诊断的诊断效能高于低年资医师独立诊断,灵敏度(88.10%比78.60%)、特异度(82.47%比57.00%)、准确率(85.47%比68.37%)和AUC(0.85比0.68)的差异均有统计学意义(P均<0.05)。结论:DL-3DUV在鉴别乳腺肿块良恶性中具有重要价值,有望成为临床辅助超声医师,尤其是低年资医师的有用工具。
