对海雷达多维图像特征融合智能检测方法

海面目标检测是雷达信号处理中的重要内容,在军事、民用领域内都有重要应用价值。在海面目标雷达信号处理中,海杂波的存在对检测算法的性能有很大影响,传统的雷达信号处理方法多基于统计理论,对于复杂环境条件和多样的目标特性检测性能下降明显。近年来深度学习技术发展迅速,为可靠的海面目标的检测方法提供了技术支持。本文对近年来目标检测算法、深度学习方法的发展进行总结,从雷达信号数据结构和维度出发,采用深度学习理论,分别提出了基于二维图像、三维视频雷达信号、多维雷达信号多通道融合的智能处理框架,并以导航雷达图像海上目标智能检测为例,提出一种Precise ROI?Faster R?CNN雷达图像检测算法,通过构建的导航雷达数据集训练和测试,相比经典恒虚警检测和Faster R?CNN检测方法有更高的检测精确度和更好的泛化能力,从而为对海雷达智能导航和目标检测提供了有效的技术途径。     

添加氨基芳香酸对铂催化烯烃硅氢加成反应的活性和选择性的影响

A series of aromatic acids has been tested as additives for the platinum-catalyzed hydrosilylation of styrene with triethoxysilane. Both excellent conversion of styrene and selectivity in favor of the β-adduct were achieved using aminobenzoic acids as additive. Moreover, the use of 4-aminobenzoic acid led to significantly superior enhancement in both catalytic activity and selectivity among the tested aminobenzoic acids. Indeed, 100% conversion of styrene and 98.4% selectivity in favor of the β-adduct were obtained. Additionally, hydrosilylations of various alkenes with a variety of platinum catalysts have also been tested, and in each case the conversion of substrate and the selectivity of the β-adduct were promoted by using 4-aminobenzoic acid as additive.     

C含量对单晶高温合金再结晶的影响

采用选晶法在真空定向凝固炉中,制备了C含量分别为0.019%、0.048%、0.074%和0.094%的单晶高温合金,合金表面吹沙后分别在1 250、1 300℃进行真空热处理,研究不同C含量对单晶高温合金再结晶的影响。结果表明:随着合金中C含量升高,碳化物含量增多,其形态由块状转变为骨架状、发达骨架状;随着C含量增加,合金再结晶层深度无明显变化趋势,这表明碳化物对合金再结晶无明显的抑制作用,随着热处理温度升高,再结晶厚层深度明显增加;C对单晶高温合金再结晶抑制作用与形成碳化物的形态和密度有关,在合金表层形成高密度的碳化物,从而对再结晶晶界形成钉扎作用,阻碍晶界的迁移,能够起到抑制再结晶的作用。