文中深入研究了贝叶斯网络推理和参数学习方法,提出了基于参数学习贝叶斯网络的目标融合识别方法,逆向分析建立了多传感器目标融合识别贝叶斯网络模型,并采用EM参数学习算法对所建立的网络模型进行参数优化。最后进行了实例仿真,仿真结果表明经过参数学习的贝叶斯网络模型比单纯依靠专家知识的贝叶斯网络模型能更加精确的识别目标。

In the paper, Bayesian inference and parameter learning were studied in-depth, target fused recognition method based on parameter learning Bayesian network was proposed, multi-sensor target fused recognition Bayesian network model was built by adverse analysis, and EM parameter learning method was adopted to optimize the model. Finally, a simulation example was given, which indicated that enemy aerial targets could be recognized more effectively by the model based on parameter learning than that based on professor knowledge.