军事信息系统中信息融合关键技术研究

发布时间：2018-01-04 07:35

本文关键词：军事信息系统中信息融合关键技术研究　出处：《电子科技大学》2016年博士论文　论文类型：学位论文

【摘要】：军事信息系统是用来在整个军事作战范围内支持指挥官筹划、决策、指挥作战等为一体的指挥控制平台,是军队战斗力的“倍增器”。它是对各军兵种所使用的信息系统进行综合设计、综合集成和综合运用,是信息系统、武器系统和军事保障系统的粘合剂。信息活动的描述、融合是军事信息系统动态要素建模、分析与仿真的基础。但是,从信息融合角度对军事信息系统的研究,当前国内外尚未见到比较系统的研究成果。本文研究的总目标是探究人脑信息融合的认知机制,研究其内在组网、特征提取、信息融合、知识表达和推理、决策等认知机理,并将研究结果应用于现代军事系统中多传感器目标跟踪与信息融合过程中。从认知科学的角度为军事信息系统应用背景下目标监测和信息融合,得出有效的融合策略提供指导或者参考。研究工作与创新点主要体现在以下四个方面:一、提出了一种基于内部模型驱动的融合进化结构体系理论方案。人的大脑处理信息过程是一种复杂的信息融合过程,其构建在一种智能、灵活的多源信息融合进化结构体系基础上。本文在借鉴人脑研究的相关理论基础上,通过对比多源信息融合进化过程与人脑的信息融合的模型、机制,指出大脑存在是由大量局部神经元构成的局部神经元回路,由此提出了一个很重要的概念:融合元;并通过研究发现多源信息融合系统的进化过程与分布式的神经网络体系结构具有很多相似特点,由此提出了一种基于内部模型驱动的融合进化结构体系理论方案。最后,从理论解析和现代军事应用背景下仿真实验证实,基于内部模型驱动的多源信息融合系统进化体系结构将大大提高作战效能和战时应变能力。二、提出了适应于战场感知信息融合的智能化分布式系统组网方法。现代战争中智能化装备的使用以及对信息处理能力的需求的不断提高,网络中心战环境逐步实现智能化组网的需求变得越来越迫切。从本质上说,网络中心战环境的战场感知,是通过有效地连接战场空间中知识实体节点将信息优势转化为战斗力,而任何一个空间中知识实体节点,都可以在条件合适的情况下产生一次信息融合事件,通过请求网络系统资源的协作,实现多传感器、智能接口的资源共享和信息交流,这使得信息之间的相关性、冗余性,以及统计分布特性等变得更加复杂和难以确定。由此,本文提出了智能化分布式系统组网方法,通过分布协作式调度网络中各融合元,融合元节点相互之间进行协调来做出决策,达到共同完成信息资源调度融合任务的目的。实验表明,分布式系统组网方法更适合融合节点获取战场信息。三、提出一种可以用于多传感器目标监测的融合非统计方法,即表决融合。适应于网络中心战环境的战场信息类型杂、信息源多、信息量巨大、信息质量差且富含诱骗,要从这样的信息环境中快速、准确、完整地提取出指挥员所需要的战场态势和环境图景来,难度极大,需要解决的问题极多。由此,提出了利用分布式网络多传感器监测目标的结果,来综合得出目标是否存在的判决技术,应用于目标身份信息融合的不确定性推理技术,提出一种简单的非统计方法,即表决融合,它可以用于多传感器目标监测的融合。实验表明:应用表决融合可获得高的节点感知性能。四、提出了一种能提高传感器组网的发现能力和连续观测能力的多传感器自适应状态融合技术。自适应状态融合技术着眼于算法间的互补性达到某种优化,算法能够根据信息源特性和目标自动选择最有效算法,具有透明性和开放性,是多层次多模型的自适应算法。实验证实:算法提出将为指挥信息系统信息融合提高精度、反隐身、反低空入侵、抗丢失目标等带来较大好处,同时可克服传感器性能不稳定等缺陷。
[Abstract]:Military information system is used to support the commander in planning military operations within the scope of the decision, combat command and control platform for the integration and military battle. It is a comprehensive design information system used in various branches of integrated and comprehensive use of the information system, weapon system and adhesive the military security system. Information fusion is the description of elements of military information system based on dynamic analysis and simulation. However, research from the angle of information fusion in military information system at home and abroad, no systematic research achievements. The overall objective of this study is to explore the cognitive mechanism of human brain information fusion, research the internal network, feature extraction, information fusion, knowledge representation and reasoning, decision-making and cognitive mechanism, the research results can be applied to modern military system in multi sensor Target tracking and information fusion process. From the perspective of cognitive science as the application background of military information system monitoring and information fusion, the effective fusion strategy to provide guidance or reference. The research work and innovations are mainly embodied in the following four aspects: first, put forward a combined evolutionary scheme of internal model driven architecture theory based on the information processing of the brain. The process is a complex process of information fusion, its construction in an intelligent, flexible multi-source information fusion system based on structure evolution. Based on the related theory of human brain research, by comparing the multi-source information fusion process of evolution and the brain information fusion model, mechanism that is, the brain local neuronal circuit composed of a large number of local neurons, thus put forward a very important concept: the fusion element; and we found The neural network architecture evolution and distributed multi-source information fusion system has many similar characteristics, this paper proposes a fusion scheme of internal structure evolution theory based on model driven. Finally, from the theoretical analysis and application background of modern military simulation results confirm that the multi-source information fusion system of internal model driven architecture evolution greatly improve the combat effectiveness and adaptability. Based on the wartime two, put forward the networking method of intelligent distributed system in battlefield information fusion. The intelligent equipment in modern war and the demand for information processing capabilities continue to improve, the network centric warfare environment gradually realize the intelligent network demand has become more urgent. In essence, the battlefield environment of network centric warfare, is the letter by effectively connecting battlefield knowledge nodes in the physical space Information superiority into combat, and knowledge entity any space node can produce an information fusion event in the right conditions, through the collaboration network request system resources, to achieve multi sensor intelligent interface, resource sharing and information exchange, which makes the correlation between information redundancy, and the statistical distribution characteristics become more complicated and difficult to determine. Thus, proposed network methods of intelligent distributed systems, distributed scheduling in the network through the fusion, fusion between the element node coordinate to make decisions, to jointly complete the information fusion task to resource scheduling. The experimental results show that the network method of distributed system suitable fusion nodes acquire the battlefield information. Three, this can be used for a multi-sensor target monitoring non statistical method, which is adapted to the voting fusion. Battlefield information environment type of network centric warfare complex, many information sources, a huge amount of information, information quality is poor and rich in deception, to fast, from this information environment accurately, completely extract the battlefield to commander and environmental picture, extremely difficult, need to solve many issues. Therefore, put forward the use of distributed multi sensor monitoring target results, to get the target existence judgment technology, applied to the target identity information fusion uncertainty reasoning technology, we proposed a simple non statistical method, namely, it can be used for voting fusion, multi-sensor target monitoring. Experimental results show that the application of voting fusion can be obtained the node sensing high performance. Four, this paper presents a multi sensor adaptive state can improve the sensor network discovery ability and continuous observation ability of fusion technology. Adaptive state finance Complementary technology focuses on the algorithms to some optimization algorithm can automatically select the most effective algorithm according to the characteristics of information source and target, transparent and open, multi-level and multi model adaptive algorithm. Experiments show that the proposed algorithm for command information system to improve the accuracy of information fusion, anti stealth, anti low intrusion the anti lost goals, bring greater benefits, and can overcome the defects such as sensor performance is not stable.

【学位授予单位】：电子科技大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TP202;E11

【参考文献】