電腦象棋深象中局程式的設計與實作

Abstract

電腦博奕在人工智慧領域中，一直是引人關注的，在深藍打敗西洋棋棋王Karsparov之後，象棋由於複雜度與西洋棋接近，在資料結構與搜尋演算法方面與西洋棋有許多相似或共通之處，成為下一個最有可能打敗人類棋王的棋類遊戲，近年來也吸引了許多人投入研究。 本論文介紹電腦象棋程式“深象”(取得了在義大利杜林舉辦之第十一屆Computer Olympiad象棋比賽銅牌)的基本資料結構、搜尋演算法與審局函數的設計，以及對所使用的資料結構與搜尋演算法進行測試與分析，並針對提高棋力的核心－審局函數，提出我們的改進方案。

關鍵字:電腦象棋、搜尋演算法、審局函數、人工智慧。

Computer chess is always deeply concerned in the field of artificial intelligence. Computer Chinese chess is most possibly the next game to defeat the top human player after DeepBlue defeated the best human chess player Kasparov. Many researchers started studying computer Chinese chess since the complexity of Chinese chess is a little higher than that of chess and both the data structure and search algorithms are similar with each other. This thesis introduces the basic data structures, search algorithms, and evaluation function of computer Chinese chess “Deep Elephant” (won the bronze medal of 11th Computer Olympiad hold in Turin, Italy) and gives the experimental results and analyses of the data structures and search algorithms it used. Finally, we propose some improvements aimed at the evaluation function－the core of strengthening the ability of the program.

Keywords : artificial intelligence, computer Chinese chess, search algorithm, evaluation function.