象棋人工智能的实现

为了实现人机对战功能，必须实现象棋的人工智能，将象棋的每个棋子都赋予一定的权重，每走一步都计算分值，选择得分最高的一步，这是象棋人工智能的基本思想。

    #ifndef AI_H#define AI_H#include "cocos2d.h"USING_NS_CC;class SceneGame;class Step : public CCObject{public:int _moveid;int _killid;int _xFrom;int _yFrom;int _xTo;int _yTo;static Step* create(int moveid, int killid, int xFrom, int yFrom, int xTo, int yTo){Step* step = new Step;step->_killid = killid;step->_moveid = moveid;step->_xFrom = xFrom;step->_xTo = xTo;step->_yFrom = yFrom;step->_yTo = yTo;step->autorelease();return step;}};class AI{public:AI(SceneGame *game);SceneGame *_game;Step *GenOneMove(int level=2);int getScore();static int _score[7];CCArray *getAllPossibleMove();void getAllPossibleMove(int idx,CCArray *arr);int getMinValue(int level,int maxScore);int getMaxValue(int level,int minScore);Step *_step;};#endif // AI_H

得分表

int AI::_score[7]=
{1000,10,10,100,50,50,20
};

创建一步

Step *AI::GenOneMove(int level)
{int maxScore=-10000;Step *ret;//find all possible access an calcute the hights scoreCCArray *possibleMOve=getAllPossibleMove();CCObject *obj;CCARRAY_FOREACH(possibleMOve,obj){Step *step=(Step*)obj;_game->fakeMove(step);int score=getMinValue(level-1,maxScore);//int score=getScore();_game->unfakeMove(step);if(score>maxScore){maxScore=score;ret=step;}}return ret;}

最大值最小值算法

int AI::getMinValue(int level,int maxScore)
{if(level ==0){return getScore();}int minScore=10000;CCArray *possibleMOve=getAllPossibleMove();CCObject *obj;CCARRAY_FOREACH(possibleMOve,obj){Step *step=(Step*)obj;_game->fakeMove(step);int score=getMaxValue(level-1,minScore);_game->unfakeMove(step);if(score<=maxScore){minScore=score;return minScore;}if(score<minScore){minScore=score;}}return minScore;
}
int AI::getMaxValue(int level,int minScore)
{if(level ==0){return getScore();}int maxScore=-10000;CCArray *possibleMOve=getAllPossibleMove();CCObject *obj;CCARRAY_FOREACH(possibleMOve,obj){Step *step=(Step*)obj;_game->fakeMove(step);int score=getMinValue(level-1,maxScore);_game->unfakeMove(step);if(score>=minScore){maxScore=score;break;}if(score>maxScore){maxScore=score;}}return maxScore;
}