基于遗传算法的CGF学习行为模型研究

通过自学习使计算机生成兵力(CGF)具有决策能力,是机器学习技术应用于军事仿真的一个重要研究方向.运用基于Agent的建模方法和学习分类器系统技术,构建了基于遗传算法的CGF学习行为模型框架,详细论述了该模型学习过程的运行周期,并将记忆功能引入CGF决策模型来加速学习进程.最后,设计了一个可视化验证系统,实验结果表明该模型的有效性和可行性.

Learning Behavioral Model of CGF based on Genetic Algorithms

The Computer Generated Force (CGF) possess decision-making skill by self-learning, which is an important research field in applying machine learning technology to military simulation. By applying modeling method based on Agent and learning classifier systems (LCSs) technology, a learning behavioral model framework of CGF was built, and the learning process of the model was discussed. Memory function was introduced into the decision model of CGF, which could improve the speed of learning. A visible validation system was realized, and the results of experiments show that this learning model is available and feasible.