遗传算法在逃逸机动策略中的应用研究

分析了基于强化学习原理和遗传算法的序贯决策规则的自动学习方法，从规划报偿和规则激活度的角度讨论和研究了规则的信度分配问题，解决了在大的状态空间中搜索和延迟评价问题，为处理复杂的决策过程提供了一种行之有效的方法。基于该方法实现了飞机的逃逸机动策略，仿真结果表明了该方法的有效性。

Learning Evasive Maneuvers Using Genetic Algorithms

The method of learning sequential decision rules automatically based on reinforcement learning and genetic algorithms is analyzed. The credit assignment of rules is discussed from the point of view of plan payoff and rule strength. Based on the method, the problem of searching in a large state space and delayed critics is solved. The research provides a new approach for dealing with complex sequential decision making problems. The evasive maneuver games for plane in combat are implemented using the genetic algorithms and reinforcement learning. Simulation results show the effectiveness of the method.