摘要
针对目前智能航迹关联算法关联准确率较低的问题,提出一种由残差网络、双向长短期记忆网络(bidirectional long short-term memory,BiLSTM)与改进的卷积注意力模块(improved convolutional block attention module,ICBAM)结合而成的残差BiLSTM-ICBAM航迹关联模型。在BiLSTM模型的基础上引入残差网络,增强模型提取航迹上下游特征的同时抑制网络退化问题;加入改进的CBAM注意力模块,分析输入信息与当前航迹特征的相关性并突出关键特征的影响,进而增强局部特征提取能力以及误差跟踪能力;在航迹关联数据上的实验结果表明,残差BiLSTM-ICBAM航迹关联模型比现有方法在准确率、稳定性中表现出了明显的性能优势。
For the current issue of low correlation accuracy in intelligent trajectory correlation algorithms,a residual BiLSTM-ICBAM trajectory correlation model is proposed,which is formed by combining a residual network,bi-directional long short-term memory network(BiLSTM),and an Improved Convolutional Block Attention Module(ICBAM).a residual network is introduced on the basis of the BiLSTM model,the model is enhanced the upstream and downstream trajectory features are extracted while network degradation problems are suppressed.The improved CBAM attention module is introduced,the correlation between input information and current trajectory features is analyzed,the impact of key features is highlighted,thereby local feature extraction capabilities and error-tracking abilities are enhanced.The experimental results on trajectory correlation data indicate that the residual BiLSTM-ICBAM trajectory correlation model exhibits significant performance advantages over the existing methods in terms of accuracy and stability.
作者
贾燎原
曹伟
张晓峰
陆翔
周恒亮
JIA Liaoyuan;CAO Wei;ZHANG Xiaofeng;LU Xiang;ZHOU Hengliang(No.724 Research Institute of China State Shipbuilding Corporation,Nanjing 210000,China)
出处
《火力与指挥控制》
北大核心
2025年第2期100-106,115,共8页
Fire Control & Command Control
关键词
航迹关联
残差网络
双向长短时记忆神经网络
卷积注意力模块
trajectory correlation
residual network
bi-directional long and short-term memory neural network
convolutional block attention module
作者简介
贾燎原(1998-),男,河南周口人,硕士。
