基于贝叶斯Noisy Or Gate网络的多传感器目标分类识别

A Model of Multiple Sensors Target Identification via Bayesian Noisy Or Gate Network

针对多传感器获取空中目标的多识别特征,提出了基于贝叶斯Noisy Or Gate网络的目标识别模型;该模型考虑未知因素的影响,将识别特征按二值节点进行网络识别结构构造,利用单个特征的识别结果,计算得到多个特征识别的任意组合,条件概率个数可以从2n减小为2n.仿真计算结果表明,该方法具有简化知识获取,节省存储空间,证据传播及时,实时性高的特点,为目标分类与识别提供了一个新的途径。

The model of target identification based on Bayesian Noisy Or Gate network was proposed using multiple identification param- eters of air target obtained by multiple sensors. The unknown factors were considered, and the Bayesian network structure was constructed using binary nodes that is derived from the identification parameters. Single result was got by single sensor target identification, and all com- bination of single sensor identification results were calculated. The computational complexity was reduced from 2n to 2n. An example was giv- en and the result demonstrate that the Noisy Or Gate network simplifies knowledge acquisition , saves storage space and allows evidence propagation in time, have a good real--time ability and high practicability.