Disconnection in the distributed heterogeneous networked unmanned weapon systems is caused by multiple weapon units＇ failure. The technical routes were analyzed to achieve resilience in the disconnection situation. A...Disconnection in the distributed heterogeneous networked unmanned weapon systems is caused by multiple weapon units＇ failure. The technical routes were analyzed to achieve resilience in the disconnection situation. A heterogeneous distributed network model of networked unmanned weapon systems was established. And an approach of adding relay weapon units was proposed to a- chieve fault tolerance after weapon units＇ failure due to attack or energy exhaustion. An improved ge- netic algorithm was proposed to determine and optimize the position of the relay weapon units. Simulation results in the MATLAB show that the improved resilience-based genetic algorithm can restore the network connection maximally when the number of relay units is limited, the network can keep on working after failure, and the implementation cost is controlled in a reasonable range.展开更多
针对区域防空反导作战中各要素复杂耦合所导致的战场态势快速演变、来袭目标数量动态变化等难题,提出一种基于可动态扩展且带时空推理的QMIX(QMIX with Dynamic extension and Spatiotemporal reasoning, QMIX-DS)的火力分配方法,以火...针对区域防空反导作战中各要素复杂耦合所导致的战场态势快速演变、来袭目标数量动态变化等难题,提出一种基于可动态扩展且带时空推理的QMIX(QMIX with Dynamic extension and Spatiotemporal reasoning, QMIX-DS)的火力分配方法,以火力单元作为智能体构建决策网络,生成火力分配策略。核心改进为:为每个智能体的决策网络设计可动态扩展特征编码模块,自适应处理数量变化的来袭目标,并引入对比学习突出目标类别属性,形成差异化特征表征;构建两层多头自注意力机制捕捉不同类别目标间的动态时空依赖关系,快速推理任务过程中的态势演变,优化火力分配策略。基于墨子平台不同规模的仿真结果表明,所提出的火力分配方法能够在动态变化的战场条件下生成有效的防空反导策略,与基线算法及其他主流算法相比,所提QMIX-DS算法在目标拦截率、阵地存活率、导弹消耗数量等指标上均体现出了优势,并在不同场景中展现出较高的扩展性和泛化性。展开更多
In order to improve weapon assignment(WA)accuracy in real scenario,an artificial neural network(ANN)model is built to calculate real-time weapon kill probabilities.Considering the WA characteristic,each input represen...In order to improve weapon assignment(WA)accuracy in real scenario,an artificial neural network(ANN)model is built to calculate real-time weapon kill probabilities.Considering the WA characteristic,each input representing one assessment index should be normalized properly.Therefore,the modified WA model is oriented from constant value to dynamic computation.Then an improved invasive weed optimization algorithm is applied to solve the WA problem.During search process,local search is used to improve the initial population,and seed reproduction is redefined to guarantee the mutation from multipoint to single point.In addition,the idea of vaccination and immune selection in biology is added into optimization process.Finally,simulation results verify the model′s rationality and effectiveness of the proposed algorithm.展开更多
装备体系组合发展规划是一项复杂的系统工程,具有重要的军事意义和研究价值。考虑到武器装备体系的关联性和复杂性,从体系的角度出发,提出一种装备体系组合发展规划方法,为装备发展论证提供思路。首先,基于装备的不同属性以及关联关系,...装备体系组合发展规划是一项复杂的系统工程,具有重要的军事意义和研究价值。考虑到武器装备体系的关联性和复杂性,从体系的角度出发,提出一种装备体系组合发展规划方法,为装备发展论证提供思路。首先,基于装备的不同属性以及关联关系,构建了装备体系异质网络模型并识别提取功能链,为后续装备发展规划建模提供基础;其次,考虑到装备发展的不确定性,建立了多阶段装备发展规划模型;然后,设计一种双重深度Q网络(double deep Q-network,Double DQN)算法来求解模型;最后,以典型装备体系发展规划为例进行演示计算,验证了所提方法的有效性和可行性。展开更多
基金Supported by the Aviation Science Foundation of China(2013ZC72006)
文摘Disconnection in the distributed heterogeneous networked unmanned weapon systems is caused by multiple weapon units＇ failure. The technical routes were analyzed to achieve resilience in the disconnection situation. A heterogeneous distributed network model of networked unmanned weapon systems was established. And an approach of adding relay weapon units was proposed to a- chieve fault tolerance after weapon units＇ failure due to attack or energy exhaustion. An improved ge- netic algorithm was proposed to determine and optimize the position of the relay weapon units. Simulation results in the MATLAB show that the improved resilience-based genetic algorithm can restore the network connection maximally when the number of relay units is limited, the network can keep on working after failure, and the implementation cost is controlled in a reasonable range.
文摘针对区域防空反导作战中各要素复杂耦合所导致的战场态势快速演变、来袭目标数量动态变化等难题,提出一种基于可动态扩展且带时空推理的QMIX(QMIX with Dynamic extension and Spatiotemporal reasoning, QMIX-DS)的火力分配方法,以火力单元作为智能体构建决策网络,生成火力分配策略。核心改进为:为每个智能体的决策网络设计可动态扩展特征编码模块,自适应处理数量变化的来袭目标,并引入对比学习突出目标类别属性,形成差异化特征表征;构建两层多头自注意力机制捕捉不同类别目标间的动态时空依赖关系,快速推理任务过程中的态势演变,优化火力分配策略。基于墨子平台不同规模的仿真结果表明,所提出的火力分配方法能够在动态变化的战场条件下生成有效的防空反导策略,与基线算法及其他主流算法相比,所提QMIX-DS算法在目标拦截率、阵地存活率、导弹消耗数量等指标上均体现出了优势,并在不同场景中展现出较高的扩展性和泛化性。
基金Supported by the National Natural Science Foundation of China(11102080,61374212)the Science and Technology on Electro-Optic Control Laboratory and Aeronautical Science Foundation of China(20135152047)
文摘In order to improve weapon assignment(WA)accuracy in real scenario,an artificial neural network(ANN)model is built to calculate real-time weapon kill probabilities.Considering the WA characteristic,each input representing one assessment index should be normalized properly.Therefore,the modified WA model is oriented from constant value to dynamic computation.Then an improved invasive weed optimization algorithm is applied to solve the WA problem.During search process,local search is used to improve the initial population,and seed reproduction is redefined to guarantee the mutation from multipoint to single point.In addition,the idea of vaccination and immune selection in biology is added into optimization process.Finally,simulation results verify the model′s rationality and effectiveness of the proposed algorithm.
文摘装备体系组合发展规划是一项复杂的系统工程,具有重要的军事意义和研究价值。考虑到武器装备体系的关联性和复杂性,从体系的角度出发,提出一种装备体系组合发展规划方法,为装备发展论证提供思路。首先,基于装备的不同属性以及关联关系,构建了装备体系异质网络模型并识别提取功能链,为后续装备发展规划建模提供基础;其次,考虑到装备发展的不确定性,建立了多阶段装备发展规划模型;然后,设计一种双重深度Q网络(double deep Q-network,Double DQN)算法来求解模型;最后,以典型装备体系发展规划为例进行演示计算,验证了所提方法的有效性和可行性。
