一种基于深度学习水下高速航行器的目标识别方法研究  

Research on an underwater high-speed vehicle target recognition method based on deep learning

作　　者：吴亚军 刘礼文 WU Yajun;LIU Liwen(Navy Armament Department,Xi'an 710077,China;The 705 Research Institute,China Shipbuilding Industry Corporation,Xi'an 710077,China)

机构地区：[1]海军装备部,陕西西安710077 [2]中国船舶集团有限公司第705研究所,陕西西安710077

出　　处：《指挥控制与仿真》2025年第2期75-82,共8页Command Control & Simulation

摘　　要：水下高速航行器作为海洋中唯一可高速航行的精确导引装备,其目标识别性能决定着任务最终完成效果。由于复杂的海洋坏境和不断升级的新型对抗器材,水下高速航行器目前面临着识别能力不足问题,急需寻找一种新型特征提取目标识别途径。本文基于深度卷积网络良好的特征挖掘能力,结合回波信号的特点,提出了一种深度学习水下目标识别模型,并利用试验场采集的数据进行了模型验证实验。同时,针对训练数据不足问题,建立了生成对抗网络进行数据集扩充,实验结果表明,本文提出的深度学习模型可有效地对水下目标进行识别,并且生成对抗网络的数据集扩充提高了模型识别准确率,为水下高速航行器智能化发展提供了新的思路。As the only high-speed navigation equipment in the ocean,the target recognition performance of the underwater high-speed vehicle determines the final completion effect of the mission.Due to the complexity of marine environment and the constantly upgrading of new countermeasure equipment,underwater high-speed vehicles are currently faced with the problem of insufficient recognition ability in complex marine environment,and it is urgent to find a new way of feature extraction and target recognition.Based on the good feature mining ability of deep convolutional networks and the characteristics of echo signals,a deep learning underwater target recognition model is proposed in this paper,and the model verification experiment is carried out by using test site data.At the same time,to solve the problem of insufficient training data,a generative adversarial networks is established to expand the data set.The experimental results show that the deep learning model proposed in this paper can effectively identify underwater targets,and the model recognition accuracy is improved by generating adversarial network data set expansion,which provides a new idea for the intelligent development of underwater high-speed vehicles.

关 键 词：水下高速航行器 目标识别 深度学习 生成对抗网络 

分 类 号：U674[交通运输工程—船舶及航道工程] TP18[交通运输工程—船舶与海洋工程]