柴油机排气噪声的有源控制研究

对四缸四冲程柴油机（4－85）1500r／min时的排气噪声的低频进行了分析利用声辐射理论,设计出引入次级声场的装置．对排气管与机器相连接法兰盘上的振动信号与噪声信号进行了相干函数分析．此外,利用高速信号处理芯片TMS320C25完成了一套自适应控制系统的初步设计．     

Impact of atmospheric turbulence on coherent beam combining for laser weapon systems

The performance of a laser weapon system based on coherent beam combining (CBC) depends on its propagation properties in the atmosphere. In this study, an analytical model based on partial coherent beam combining (PCBC) for assumed coherence coefficients between beams in a CBC lattice was developed. The Kolmogorov model of atmospheric turbulence and the Hufnagel-Valley model of Cn2 dependence on atmospheric parameters were implemented. Novel simplified metrics were proposed to assess the CBC performance. Several beam profiles (super-Gaussian, truncated Gaussian, etc.) and ge-ometries were analyzed in terms of maximal intensity in the far field. An approximate formula for PCBC efficiency dependent on the Fried radius was proposed. The results of CBC modeling were compared to those of the Gaussian beam propagation model in a turbulent atmosphere. The dependence of CBC performance on the Cn2 parameter, range, and elevation angle was analyzed. It could be concluded that the application of CBC for medium and long range propagation is impractical without an effective adaptive optics system.     

用于视频图像帧间运动补偿的深度卷积神经网络

为探索深度学习理论在视频图像帧间运动补偿问题中的应用,提出一种用于视频图像帧间运动补偿的深度卷积神经网络。该网络由卷积模块和反卷积模块构成,可以处理不同分辨率输入图像并具备保持较完整图像细节的能力。利用具有时序一致性的视频图像序列构造训练样本,采用随机梯度下降法对设计的深度卷积神经网络进行训练。视觉效果和数值评估实验表明,训练得到的网络较传统方法能更有效地进行视频图像帧间运动补偿。