舰船通信关联信息目标数据检索方法研究
Research on target data retrieval methods for ship communication association information
李中 1郭云丽1
作者信息
- 1. 江苏海事职业技术学院,江苏南京 211170
- 折叠
摘要
舰船通信系统会产生大量的通信关联信息,难以在海量的通信关联信息中,捕获舰船通信数据的关键属性,对此,提出舰船通信关联信息目标数据检索方法.通过计算舰船通信关联信息全局检索的特征量,对舰船网络数据进行聚类和信息融合,结合深度卷积神经网络和哈希学习算法构建目标数据检索模型,得到目标数据的初步检索结果.使用汉明距离计算初步检索结果的相似度,比较目标检索数据与初步检索结果的哈希码之间的汉明距离,输出与舰船通信关联信息目标数据相似度最高的检索结果,实现目标数据检索.实验验证,该方法实现的目标数据检索MAP值能够达到98％以上,实现的不同信息种类数据检索的时间消耗均能保证在在8 ms以下,能够为相关人员提供更加便捷和可靠的信息服务,为舰船管理、维护和作战决策提供有力的支持.
Abstract
The ship communication system generates a large amount of communication related information,making it difficult to capture the key attributes of ship communication data in a massive amount of communication related information.Therefore,a target data retrieval method for ship communication related information is proposed.By calculating the feature quantity of global retrieval of ship communication correlation information,clustering and information fusion of ship net-work data are carried out,and a target data retrieval model is constructed by combining deep convolutional neural networks and hash learning algorithms to obtain preliminary retrieval results of the target data.Calculate the similarity of preliminary search results using Hamming distance,compare the Hamming distance between the target search data and the hash code of the preliminary search results,output the search result with the highest similarity to the target data related to ship communic-ation,and achieve target data retrieval.Experimental verification shows that the target data retrieval MAP value achieved by this method can reach over 98％,and the time consumption for data retrieval of different types of information can be guaran-teed to be below 8 ms.It can provide more convenient and reliable information services for relevant personnel,and provide strong support for ship management,maintenance,and combat decision-making.
关键词
舰船通信/关联信息/目标数据检索/深度卷积神经网络Key words
ship communication/related information/target data retrieval/deep convolutional neural network引用本文复制引用
出版年
2024
NETL
NSTL
万方数据