为使概率逻辑的不确定性推理方法能应用于较大规模的知识库，本文基于一个实际专家系统知识库的开发经验，在概率逻辑公式一致性区间的一般算法基础上，为概率逻辑公式集的分解设计了一种合并聚类算法．对于不同背景的概率逻辑知识库，只要公式集具有一定的分层结构性质，该算法就能保证Ｄａｎｔｚｉｇ－Ｗｏｌｆｅ分解的联合计算模型适用于概率逻辑推理．测试结果表明，该算法对于数１０个变量和子句的实例可收到很好的效果． In order to make the method of uncertainty reasoning of probability logic applicable to a large-scale knowledge base, based on the development experience of a real expert system knowledge base, based on the general algorithm of consistency interval of probability logic formula, The decomposition of the design of a combined clustering algorithm. For the probabilistic logic knowledge bases of different backgrounds, the algorithm can ensure that the joint computing model of Dantzig-Wolfe decomposition is suitable for probabilistic logic reasoning, as long as the formula set has a certain hierarchical structure. The test results show that this algorithm can be very effective for the instances of several 10 variables and clauses.