Abstract: This paper defines some new concepts, such as atom plan, independent plan, child plan and father plan etc. Based on these concepts, it proposes a novel algorithm for plan analysis in the graphplan framework. The algorithm provides a solution for the problem in which the initial preconditions change after getting a valid plan. It proposes correlative actions, correlative plans and the algorithm for judging correlative plans, sequentially provides the method for plan synthesis, synthesizing two correlative plans will save resource cost in the real world.

Key words: Graphplan, Plan decomposition, Plan synthesis, Atom plan, Correlative plan

摘要： 提出了原子规划、独立规划、子规划以及父规划等概念，给出了图规划框架下的规划分解算法，有效地解决了在现实世界中求出有效规划后，初始条件发生改变的情况。定义了相关动作、相关规划以及相关规划的判定方法，提出了规划合成算法。几个有效规划的单独执行会浪费资源，针对这种情况该文提出了解决方案。

关键词: 图规划, 规划的分解, 规划的合成, 原子规划, 相关规划