【作者】 钱斌；

【导师】 姜长生；

【作者基本信息】 南京航空航天大学 ， 武器系统与运用工程， 2007， 硕士

【摘要】 所谓直升机智能火控系统,就是将人工智能技术与武装直升机火控系统相结合,利用人工智能技术将机载武器火力指挥控制问题进行智能化设计,并以神经网络、模糊控制、遗传算法等为实现基础,以对付现代直升机空战日趋严峻的不确定性和复杂性。本文所设计的基于数据链的直升机智能火力控制系统,主要围绕四个部分内容展开:即直升机导弹三维智能导引律、直升机机间数据链系统、直升机空战决策系统,三维智能导引律下的导弹攻击仿真研究。首先,参照第三代短波通信技术系统MTL-STD-188-141B,分析了影响数据链通讯效果的因素,然后针对这些因素提出了一种直升机机间数据链的设计方案,并用卡尔曼滤波算法对数据链传输过程中的延迟进行了补偿,最后用大量的数字仿真验证了该算法的有效性。其次,进行了基于数据链的直升机空战协同决策系统的研究:首先给出了直升机空战态势分析方法;然后在该态势分析方法的基础上,给出了传统的直升机空战威胁判断及编队方法并且用遗传算法来对空战决策进行了优化;接着,研究了基于蚁群算法的火力分配方法;最后,进行了直升机空战机动决策的研究,既分析了传统编队战术机动,又在此基础上提出了一种基于补偿模糊神经网络的直升机机动决策方法。以上的所有方法均进行了大量的数字仿真,仿真结果也证明了以上诸方法的正确性与有效性。在本文的接下来一章中用各种数据指标来分析了导弹比例导引律的优缺点,然后提出了一种基于遗传退火算法的变论域三维模糊导引律,同时给出了一种用RBF网络计算导弹三维攻击区的方法,并且通过数字仿真对两者性能进行了比较。最后一部分阐述了在VC++环境中使用Multigen-Paradigm creator和Vega实现三维动画仿真的方法,给出了我方直升机发现目标,判断目标并做出决策后发射导弹攻击目标的空战三维动画方法,并且用三维动画形象地演示了导弹攻击目标的全过程,生动直观地验证了导引律的正确性。更多还原

【Abstract】 The Intelligent Fire Control System of Helicopters is a system which combines Artificial Intelligence (AI) technology and Helicopter Fire Control. Fuzzy Control, Neural Networks and Genetic Algorithm(GA) are used in this system. This paper mainly works on air-to-air tactics data link, cooperative decision-making of multi-aircrafts attacking multi-targets, and three-dimension intelligent guidance law for air-to-air missile.Part I, A scheme of air-to-air tactics data link system is brought forward , and then the delay of the data translation is compensated using Kalman filtering algorithms.Part II, this part mainly works on the decision-making system of helicopter cooperative air-combat. At first, a method of situation analyzing is devised; Then a means of traditional helicopter air-combat formation is proposed,and we use the GA to optimize the formation; Then we propose a method of weapon distribution using Ant Colony Algorithm. At last, this paper mainly works on the maneuver decision of helicopter air-to-air combat, not only the formation maneuver based on traditional method, but also the respective maneuver in a formation based on compensative Fuzzy-Neural networks. All of above are simulated with MATLAB and C language and proved correct.Part III, PNG and a self-adapt region fuzzy logic guidance based on Genetic Algorithm and Simulated Annealing are devised. And then a method of launch envelope caculating using RBF networks is imposed, these methods are all simulated with MATLAB.In the end, the method of using Multigen-Paradigm creator and Vega in Visual C++ environment is discussed. And a 3D animation simulation software is developed to show the process of air-to-air missile attacking target.更多还原