基于计算机博弈的五子棋算法研究

发布时间：2018-11-07 08:43

【摘要】：人工智能是近年来最热的话题,各种人工智能产品层出不穷。作为它分支的机器博弈也很受重视。近年来国家也一直很重视这一领域,支持推广了很多有关计算机博弈的比赛。机器博弈是通过模拟人类智慧去解决实际问题。这也是研究的实际应用价值。本设计采用棋类中的五子棋为研究对象,对现有研究成果进行研究,综合国际前沿研究趋势,对博弈过程中搜索算法进行主要研究设计。下面为主要的设计工作内容。首先,对计算机博弈算法进行主要研究和学习。了解和研究五子棋发展状况、下棋规律和规则。对整体系统进行框架设计和棋局生成、走棋显示、时间计时等的界面设计。其次,根据五子棋下棋规则,提出本文拟采用的博弈树搜索算法并实现。以此为基础进行算法改进。通过在Alpha-Beta剪枝算法中引入迭代加深以及窗口搜索的方法,对比初始程序有很大的提升。评估函数对整个系统也起着重要的作用,基于此设计中又对评估函数进行了研究和设计。最后,为了大幅度提高系统的棋艺,针对搜索效果不理想的问题,提出了使用机器学习的方法来代替采用的搜索算法的方法。本文设计出了成型可以进行独立下棋的程序环境,可供独立完成博弈过程。并且实验效果证明棋艺有很大的提高,具有一定的研究意义和使用价值。
[Abstract]:Artificial intelligence is the hottest topic in recent years, a variety of artificial intelligence products emerge in endlessly. The machine game, which is a branch of it, is also valued. In recent years, the country has also attached great importance to this field, supporting the promotion of a lot of computer games. Machine game is to simulate human intelligence to solve practical problems. This is also the practical application value of the study. This design takes Gobang in chess as the research object, studies the existing research results, synthesizes the international frontier research trend, carries on the main research design to the search algorithm in the game process. The following is the main design work. Firstly, the computer game algorithm is studied and studied. Understand and study the development of Gobang, chess rules and rules. Frame design and chess game generation, chess display, time-timing interface design for the whole system. Secondly, according to the rules of Gobang, the game tree search algorithm proposed in this paper is implemented. Based on this, the algorithm is improved. The method of iterative deepening and window searching is introduced into Alpha-Beta pruning algorithm, which is much better than the initial program. Evaluation function also plays an important role in the whole system. Finally, in order to improve the chess skill of the system greatly, aiming at the problem that the search effect is not ideal, the method of using machine learning to replace the search algorithm is put forward. In this paper, we design a program environment in which we can play chess independently, which can be used to complete the game independently. And experimental results show that chess has a great improvement, and has a certain significance and use value.
【学位授予单位】：沈阳理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP18

【参考文献】