飞行器表面等离子体融合分析法  

Aircraft surface plasma fusion analysis method

作　　者：李昀桓 李奇 金魁 Li Yunhuan;Li Qi;Jin Kui(Faculty of Electronic Information Engineering,Huaiyin Institute of Technology,Huaian 223022,China)

机构地区：[1]淮阴工学院电子信息工程学院,江苏淮安223022

出　　处：《南京理工大学学报》2023年第6期851-858,共8页Journal of Nanjing University of Science and Technology

摘　　要：为对飞行器表面的等离子体态进行实时监测和分析,提出了一种新型的融合神经网络的等离子体分析方法。对电磁波在等离子体中的传播情况进行了分析,使用渐近薄膜近似和初等矩阵运算的方法计算相关参数,再引入神经网络进行多特征的优化,并使用自适应学习率调整。实验结果表明:在数据传输稳定性测试中,该文方法在碰撞频率为0.5 GHz时的数据传输速率在1020 Mbps附近波动;进行净运算时间测试时,在碰撞频率为5 GHz、面积区域为2 m^(2)时的净运算时间为253 ms。To monitor and analyze the plasma state on the surface of the aircraft in real-time,a novel plasma analysis method integrating neural networks is proposed.In the process,the propagation of electromagnetic wave in the plasma is analyzed,and then the relevant parameters are calculated by means of asymptotic film approximation and elementary matrix operation.Then,the neural network is introduced for multi feature optimization,and adaptive learning rate adjustment is used.The experimental results show that in the stability testing of data transmission,the data transmission bit rate fluctuates around 1020 Mbps at a collision frequency of 0.5 GHz;when conducting net computation time testing,the research method shows a net computation time of 253 ms at a collision frequency of 5 GHz and an area of 2 m^(2).

关 键 词：电磁波 神经网络 等离子体 渐近薄膜近似 初等矩阵运算 自适应学习率 数据传输速率 净运算时间 

分 类 号：P352.3[天文地球—空间物理学]