变结构离散动态贝叶斯网络及其推理算法

目前的动态贝叶斯网络的研究,是定义在每一个时间片的静态贝叶斯网络结构和参数都一致的基础上,对于过程突变,参数变化等情况就难以适应.为了解决这个问题,提出变结构离散动态贝叶斯网络的概念,并根据概率和动态贝叶斯网络的理论,推导出变结构离散动态贝叶斯网络的推理方法,对算法进行了验证并结合环境变化时的路径选择问题,进行了计算仿真.计算和仿真结果证明了文章提出的变结构离散动态贝叶斯网络的概念和推理算法的正确性.

Structure varied discrete dynamic Bayesian network and its inference algorithm

Till now,the research on dynamic Bayesian networks is based on such a assumption that all the static Bayesian networks in all time chips are the same, not only in the architecture,but also in the parameters.This kind of dynamic Bayesian network has such drawbacks that it is difficult to deal with the process whose architecture and parameters are varying.In order to cope with such a problem,we propose a concept of structure varied dynamic Bayesian network and deduce an inference algorithm of the structure varied dynamic Bayesian network based on the strict probably theory,and verified the inference algorithm by an example.Combined with the route selection problem,we did a simulation using the conception and inference algorithm mentioned above.The simulation result implies that the conception and inference algorithm of structure varied dynamic Bayesian network are both correct.