基于LSTM运动模型分类的水下目标跟踪算法分析  

Analysis of Underwater Target Tracking Algorithm Based on LSTM Motion Model Classification

作　　者：刘承林 LIU Chenglin(Shenyang Ligong University,Shenyang,Liaoning 110159,China.)

机构地区：[1]沈阳理工大学,辽宁110159

出　　处：《电子技术(上海)》2025年第1期53-55,共3页Electronic Technology

摘　　要：阐述针对跟踪水下移动目标时,运动模型区分度不足导致跟踪效果不佳的问题,提出一种基于LSTM运动模型分类的目标跟踪算法,以提高目标跟踪的精度。仿真实验表明,与IMM-BP算法相比,IMM-LSTM-C算法的精度提升了2.7%~7.1%。This paper describes a target tracking algorithm based on LsTM motion model classification(Interactive Multi-Modeling using Long and Short-Term Memory Classification,IMM-LSTM-C),which enhances the accuracy of target tracking.in response to the issue of insufficient differentiation of motion models when tracking underwater moving targets,which leads to suboptimal tracking performance.Simulation experiments demonstrate that,compared to the IMM-BP algorithm,the IMM-LSTM-C algorithm achieves an accuracy improvement ranging from 2.7%to 7.1%.

关 键 词：水下传感器网络 目标跟踪 运动模型分类 IMM-LSTM-C 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置] TP391.44[自动化与计算机技术—控制科学与工程] TN29.5[电子电信—物理电子学] TP18