期刊文献+
共找到2篇文章
< 1 >
每页显示 20 50 100
基于深度卷积神经网络的多飞行器构型保持研究
1
作者 槐泽鹏 王洪波 龚旻 《兵器装备工程学报》 CAS CSCD 北大核心 2021年第2期15-22,共8页
基于深度卷积神经网络设计了“智能-校正”算法以快速生成制导指令;深度学习模型是由卷积层、全连接层、批量归一化层和残差网络(ResNet)组合成的一个19层深度神经网络,飞行器状态变量转化为类似于图像像素矩阵的二维矩阵,通过sgd优化... 基于深度卷积神经网络设计了“智能-校正”算法以快速生成制导指令;深度学习模型是由卷积层、全连接层、批量归一化层和残差网络(ResNet)组合成的一个19层深度神经网络,飞行器状态变量转化为类似于图像像素矩阵的二维矩阵,通过sgd优化和权重衰减正则项将神经网络的训练误差降低至约0.025°;之后针对飞行器运动存在制导指令泛化误差累积的问题又设计了校正环节;最后经多组仿真证明了“智能-校正”算法兼顾了速度和精度,不仅成功实现了构型保持,而且将解算速度提升了约6倍,同时解决了泛化误差累积带来的影响。 展开更多
关键词 深度卷积神经网络 多飞行器协同 构型保持 智能-校正
下载PDF
Design and Implementation of High Power Factor LED Driver with Hot Swap, Smart Output Voltage Regulation and Dimming Control
2
作者 Yu-En Wu Kuo-Chan Huang 《Journal of Energy and Power Engineering》 2013年第12期2354-2362,共9页
In this paper, a high power factor LED driver with hot swap, smart output voltage regulation and dimming control is proposed. The dimming control is used to change LED brightness. During converter is working, the hot ... In this paper, a high power factor LED driver with hot swap, smart output voltage regulation and dimming control is proposed. The dimming control is used to change LED brightness. During converter is working, the hot swap function supply users to remove and insert LED module. The smart output voltage can regulate quickly and rightly output voltage in different number of LED series connection. The system consists two stages, one is 50 W flyback converter which is used as power factor corrector, it is input source is 110-220 V, PF (power factor) is about 0,994. The other is Boost DC/DC converter, it can offer 35-60 V of output voltage. Finally, a prototype has been built and tested. The simulation and experimental results are shown to verify the feasibility of the proposed method. 展开更多
关键词 Hot swap smart output voltage regulation dimming control flyback converter boost converter.
下载PDF
上一页 1 下一页 到第
使用帮助 返回顶部