基于循环神经网络的弹药着速预测  

Prediction of Ammunition Target Velocity Based on Recurrent Neural Network

作　　者：王现磊 王义江 陈春江 吴家健 WANG Xianlei;WANG Yijiang;CHEN Chunjiang;WU Jiajian(Unit 63861 of PLA,Baicheng 137001,China)

机构地区：[1]解放军63861部队,吉林白城137001

出　　处：《火力与指挥控制》2024年第3期119-123,共5页Fire Control & Command Control

摘　　要：针对当前立靶对雷达回波信号产生干扰,对着靶速度的提取过多依赖信号的滤波、特征点提取以及人工经验等问题,提出了利用循环神经网络(recurrent neural network,RNN)来分析弹药飞行的速度规律。通过某型穿甲弹速度数据的训练集和测试集,来进行弹药着速的预测。通过对比随机森林回归(random forest regression,RFR)算法和基于BP的多层感知机(multi-layer perceptron,MLP)模型在测试集上的均方根误差(root mean square error,RMSE),RNN收敛速度较快且在RMSE上的误差小,对于弹药的着靶速度有着很好的预测效果。In response to the current problems that vertical target interferes with radar echo signals and the extraction of the target velocity relies too much on signal filtering,feature point extraction and manual experience,etc.,a recurrent neural network is proposed to analyze the velocity law of ammunition flight.Through the training set and test set of an armor-piercing ammunition velocity data,the ammunition target velocity is predicted.Root Mean Square Error of RFR algorithm and the BPbased MLP model in the test set are compared,the RNN converges faster and has a smaller error on the RMSE,which has a good prediction effect on the ammunition target velocity.

关 键 词：着靶速度 RNN 模型对比 RMSE 

分 类 号：TJ410[兵器科学与技术—火炮、自动武器与弹药工程]