为了提高高密度航天发射任务条件下的雷达维修能力,缩短维修时间,研究了一种基于贝叶斯网络的雷达故障远程诊断系统。设计了该故障诊断系统的硬件总体结构和软件功能架构,介绍了雷达故障数据的采集和加密传输过程。利用贝叶斯网络建立雷达故障模型,并通过期望最大算法确定了故障模型的参数;采用联合树算法,实现故障定位推理。故障诊断实例表明:与传统的故障诊断方法相比,基于贝叶斯网络的故障诊断方法能够有效识别雷达的故障类型,具有故障定位准确、运行时间短等优点。

In order to improve the ability of radar maintenance on the condition of high density aerospace launch mission, and shorten repair time, a radar fault remote diagnosis system based on Bayesian network is discussed. The hardware overall system architecture and the software function framework of the fault diagnosis system are designed. The acquisition and encryption transmission of fault data are introduced. The radar fault model is established by adopting Bayesian network, and the parameters of the fault model are confirmed through the expectation maximum algorithm, and the fault location reasoning is realized by adopting junction tree algorithm. Through the example of fault diagnosis, the result demonstrates that compared with the traditional fault diagnosis method, the fault diagnosis method based on Bayesian network can identify effectively the fault type, and have the advantages of fault location accuracy and less operation time, etc.

TN919.5

四川省教育厅青年基金资助项目