为解决贝叶斯网络(BN)的小样本建模和基于知识建模下的网络不准确问题,结合网络节点间互信息,本文提出了基于互信息的BN动态优化算法。该算法基于贪婪算法,通过引入节点资源消耗概念,选择在最大资源消耗上限下能够获取最大信息的节点组合,从而实现对网络的优化。仿真结果表明,所提出的优化算法能够有效地提高推理准确度,同时大幅度降低资源和时间消耗。 In order to solve the small sample modeling of Bayesian Networks (BN) and the network inaccuracy based on knowledge modeling, combined with the mutual information between network nodes, a BN dynamic optimization algorithm based on mutual information is proposed in this paper. Based on the greedy algorithm, this algorithm introduces the concept of node resource consumption and chooses the node combination which can obtain the maximum information under the maximum resource consumption ceiling, so as to optimize the network. The simulation results show that the proposed optimization algorithm can effectively improve the accuracy of inference, while significantly reducing resource and time consumption.