摘要
应用神经网络技术对寻的式制导武器常用的PCM码型进行了识别,以验证这种解码方法的有效性。通过建立线性神经网络,采用训练向量长度优先和数量优先两种方式对网络进行训练,并对下一激光脉冲信号的产生时刻进行了预测。仿真结果表明,线性神经网络在2.3个周期内就可以准确地预测,并且脉冲在时间轴上小范围的抖动不能影响网络的预测精度,这表明神经网络技术可以有效地对脉冲编码进行解码操作,具有一定的工程实用性。
Neural network technology is used to recognize PCM code of homing guidance weapon,and this can validate the method of decoding.Firstly,through building linear neural network and train network using two mode,that are size priority and quantity priority of train vector,the generation time of next laser pulse was forecasting.The result of simulation showed the PCM code can be recognized correctly if 2.3 periodic times was offered,and laser pulse whipping on the time axis couldn′t affect the precision of forecasting.This indicated that neural network technology can decode laser pulse code,and this method is suitable for engineering.
出处
《激光与红外》
CAS
CSCD
北大核心
2010年第10期1076-1079,共4页
Laser & Infrared
关键词
激光编码
神经网络
解码
laser pulse-coded signal
neural network
decoding
