摘要
战斗机飞行员海上遇险搜救的核心是获取遇险定位信息,由于受海上气象、浪涌等多种因素影响,手持和穿戴式救生终端选择恰当的时机快速发起遇险位置报告的相关能力较弱。以飞行员某装备北斗RDSS短报文有效发射为研究对象,利用深度神经网络,提出一种海上环境下发射时机学习预测模型,将高度、经纬度、方向、俯仰角、电量、速度、加速度等参数转换为单精度浮点数输入深度神经网络,经过10个隐层,每层32个神经元,得到通信成功率预测和发射延迟预测结果,在某装备工作时长延长43%的基础上,短报文发射成功率提高1倍。
When the pilot is in maritime distress,the most important thing about rescue is to obtain distress location information.However,due to the complexity of the weather and conditions,The pilot’s hand-held life-saving communications equipment has become less effective,and power consumption has also increased significantly.This paper studies the application of short message communication in Beidou satellite navigation system at sea.Based on the deep neural network,this paper proposes a prediction model which converts the parameters of height,latitude and longitude,direction,pitch angle,electric quantity,velocity and acceleration into single precision floating-point input depth neural network.The results of communication success rate prediction and transmission delay prediction can be obtained by processing the data in 10 layers and 32 neurons.The test shows that the success rate of message launch is doubled,on the basis of the 43%extension of the working time.
作者
杨航
李洪烈
方芳
程春华
Yang Hang;Li Honglie;Fang Fang;Cheng Chunhua(Naval Air University Qingdao Campus,Qingdao,Shandong 266041,China;Naval Qingdao Special Service Convalescent Center,Qingdao,Shandong 266000,China)
出处
《信号处理》
CSCD
北大核心
2020年第12期2061-2066,共6页
Journal of Signal Processing
关键词
深度学习
航空遇险
RDSS
预测模型
复杂海况
deep learning
aviation distress
RDSS
prediction model
complex sea conditions
