摘要
如何进行贴近实际的合成火力分配是一个极具挑战性的课题。考虑到实际作战任务的多阶段特性,提出了一种新的回合制对抗策略及博弈模型,并设计了高效的确切解算法。红方对抗策略的首要目标设计为每一回合均最大化蓝方的损失,且最小化己方的损失。其次,针对战术目标的优先级问题,引入了权重系数分别表示蓝方和红方不同兵力的重要性,其值越高代表其被打击或保护的优先级越高。此外,首次引入了表征红方对自身损失敏感程度的超参数,其值小即对损失不敏感,意味着可采取较激进的战术。针对相应的整数线性规划模型,设计了分支定价算法。实验结果不仅表明了数学建模的正确性与合理性,而且验证了分支定价法相较于遗传算法的优越性。对抗策略设计、优先级量化方案以及损失敏感程度度量的合理性也得到了充分验证。此外,该新颖的对抗博弈模型及算法部分表现出了与传统兰彻斯特作战模型的一致性;而可调节的损失敏感参数,使得模型具有较好的兵力分配的灵活性,在达成战役目的的前提下,可实现兵力的阶段性投入并避免某一阶段无某一类型兵力可用的情况。
How to perform realistic synthetic firepower distribution is a very challenging topic.Considering the multi-stage characteristics of actual combat tasks,this paper proposes a new turn-based confrontationstrategy and game model,and designs an efficient exact solution algorithm.The primary goal of the red force confrontation strategy is to maximize the blue force losses and minimize the red force losses on each turn.Second,in terms of the issue of prioritization of tactical objectives,we introduce weight coefficients to represent the importance of the blue and red forces,respectively.The higher the weight coefficient,the more it should be attacked or protected.In addition,we introduce a hyperparameter to represent the sensitivity of the red force to its own losses in the combat.The smaller of the hyperparameter value,the less sensitive it is to losses,i.e.,it can be more tactically aggressive.The experiment results not only show the correctness and rationality of the mathematical modeling,but also verify that the branch pricing algorithm is superior to the genetic algorithm.The rationality of the confrontation strategy design,the prioritization quantification scheme,and the loss sensitivity measure are validated.In addition,the novel antagonistic game model and the algorithm exhibit consistency with the traditional lanchester combat model.The adjustable loss-sensitive parameter also allows the model to have better flexibility in force distribution,and allowing for the phased commitment of forces and avoiding the situation where no certain type of force is available at a certain stage while achieving the campaign objectives.
作者
王春光
许乐
蔡立言
徐勇
WANG Chunguang;XU Le;CAI Liyan;XU Yong(Unit 31002,Beijing 100041,China;School of Computer Science and Technology,Harbin Institute of Technology(Shenzhen),Shenzhen,Guangdong 518055,China)
出处
《兵器装备工程学报》
CAS
CSCD
北大核心
2023年第7期178-188,共11页
Journal of Ordnance Equipment Engineering
基金
国家自然科学基金项目(61876051)。
关键词
合成火力分配
回合制对抗策略
博弈模型
整数线性规划模型
分支定价法
synthetic firepower distribution
turn-based confrontation strategy
game model
integer liner programming model
branch pricing algorithm
