动态贝叶斯网络在战术态势估计中的应用*

针对战术态势估计的特点和要求，分析和建立了应用于态势估计的动态贝叶斯网络模型。该模型以离散变量集为研究对象。由于该动态贝叶斯网络满足Markovian特性和平稳特性，降低了网络的复杂度。相比较于贝叶斯网络模型，该动态贝叶斯网络模型考虑了时序因素，将前时刻的态势因素作为当前时刻态势估计的证据的一部分，并能对下一时刻的态势进行预测。文中采用集树（junction tree）算法，利用相关的贝叶斯网络推理软件进行了实验，实验结果表明基于动态贝叶斯网络的估计结果较贝叶斯网络的估计结果好，验证了该模型的有效性。

Application of dynamic Bayesian networks in tactical situation assessment

A dynamic Bayesian networks model for situation assessment is constructed in this paper. The model considers discrete variable set as the object of study. Since the dynamic Bayesian networks model satisfy the Markovian property and stationary, the complexity of the networks is reduced. Compared with Bayesian networks , this dynamic Bayesian networks model takes temporal factor into account . Through this model, the current situation assessment is not only associated with the evidence at the moment ,but also with previous situation . Simultaneously, the next time situation can be predicted . The junction tree algorithm and Bayesian inferencing software is used in the experiments . Experimental results show that the situation assessment result is better than that of Bayesian networks and that the dynamic Bayesian networks model is effective.