航空科学技术2025，Vol.36Issue(1)：39-45,7.DOI:10.19452/j.issn1007-5453.2025.01.005
基于虚拟样本生成的脱靶量预测方法研究
Research on Miss Distance Prediction Methods Based on Virtual Sample Generation
摘要
Abstract
Accurately predicting the miss distance can enhance the accuracy of weapon system strikes and improve the reliability of the system.Due to the traditional miss distance prediction methods relying on the accuracy of recursive models,this paper proposes a long short-term memory(LSTM)network off target prediction method based on virtual sample generation.Based on the analysis on the characteristics of off target time series,an error source model is established using its error source characteristics,and combined with its characteristics to form a virtual sample.Subsequently,the LSTM network was used to learn the nonlinear mapping relationship of off target distance,forming an LSTM off target prediction model based on virtual samples.Finally,the off target prediction model was validated using measured data and compared and analyzed with the prediction results based on the novel Kalman filter algorithm and the least squares method.The experimental results show that the LSTM off target prediction method based on virtual sample generation can improve prediction accuracy by supplementing virtual samples in the case of insufficient data,and has good application prospects.关键词
脱靶量预测/虚拟样本生成/长短期记忆神经网络/脱靶量误差模型/时间序列Key words
miss distance prediction/virtual sample generation/LSTM neural network/miss distance error model/time series分类
信息技术与安全科学引用本文复制引用
赵保琛,何山,吴盘龙,王轲,陈伟..基于虚拟样本生成的脱靶量预测方法研究[J].航空科学技术,2025,36(1):39-45,7.基金项目
航空科学基金(2022Z037059001,20220001059001) （2022Z037059001,20220001059001）
江苏省卓越博士后计划(JB23147) Aeronautical Science Foundation of China(2022Z037059001,20220001059001) （JB23147）
Jiangsu Funding Program for Excellent Postdoctoral Talent(JB23147) （JB23147）
