提出以2个 Player 组成的一个博弈系统.每个 Player 代表1个智能体,每个 Player 的输入是另一个 Play-er 的输出,每个 Player 的行为由神经元网络来描述,按照各自目标函数来调整其权值及给出策略.另外,针对2个Player 的博弈,基于心理学理论,首次给出15种博弈行为.用神经元网络来模拟每个 Player 思维判断,使各种博弈行为更接近于人的行为.通过例子仿真,结果表明提出方法是可行有效的. A game system consisting of two players is proposed. Each player represents an agent, and each player’s input is another Play-er’s output. Each player’s behavior is described by a neural network. According to their respective objective functions To adjust its weight and give the strategy.In addition, for the two players of the game, based on the theory of psychology for the first time given 15 kinds of game behavior.Using the neural network to simulate each Player thinking to judge, so that a variety of game behavior more Close to the behavior of human.Through the example simulation, the results show that the proposed method is feasible and effective.