摘要
针对复杂战场环境下的无人机决策推理模型参数学习的问题,提出基于遗传算法的离散动态贝叶斯网络参数学习算法。该算法将模型网络参数的最大似然估计函数作为遗传算法的适应度函数,在全局进行并行优化搜索,得到最优的网络参数,从而提高了决策推理模型对复杂环境的快速适应。仿真结果表明,该算法可以获得准确的离散动态贝叶斯网络参数,能够有效地解决复杂战场环境下无人机威胁评估问题,为无人机的自主任务决策提供有效的参数保障。
A parameter learning algorithm for the discrete dynamic Bayesian networks based on genetic algorithm is proposed concerning the problem of the parameter learning for the UAV＇s decision-making and reasoning model under complicated tactical environment.The algorithm employs the maximum likelihood estimating function of the model＇s network parameters as the fitness function for the genetic algorithm and searches the global optimum.By virtue of these measures,the algorithm＇s quick adaptability to the complicated environment is achieved.Furthermore,the simulation study also demonstrates that the algorithm can find the precise parameters for the discrete dynamic Bayesian networks,which underlies the solid parameter support for the autonomous decision making needed by UAV＇s task assignment.
出处
《火力与指挥控制》
CSCD
北大核心
2013年第1期26-29,共4页
Fire Control & Command Control
基金
全国高校博士点基金资助项目(20116102110026)
关键词
离散动态贝叶斯网络
参数学习
遗传算法
无人机
决策推理模型
discrete dynamic Bayesian network
parameter learning
genetic algorithm
Unmanned Aerial Vehicle（UAV）
model of decision-making and reasoning
