基于贝叶斯网络推理的起落架系统故障诊断技术研究

Fault Diagnosis of Landing Gear System Based on Bayesian Network Inference

摘要:

民机起落架系统结构复杂,是典型的故障多发系统,实际诊断过程主要依赖于排故手册流程和工程经验积累,存在诸多不确定性因素。贝叶斯网络是用有向无环图的形式表达变量间因果关联关系,可以充分利用专家知识和试验信息进行基于概率的统计推断,适于处理复杂系统的不确定性问题。通过深入分析某型民机起落架技术资料,建立了基于贝叶斯网络的起落架系统诊断架构,结合专家知识和维护经验提出了基于贝叶斯网络的起落架系统故障诊断方法,并给出了网络推理流程,提升了起落架系统故障诊断效率和精度。

Abstract:

Civil aircraft landing gear system is a typical complex system with frequent fault occurrence. Practical fault diagnosis mainly relies on Trouble-Shooting Manual and engineering experience, there exists large uncertainty. Bayesian network is a directed acyclic graph which can express the casual correlation between variables. It makes full use of experts’ knowledge and test information for probabilistic inference, which is suitable to address uncertainty problems. A diagnostic framework of the landing gear system is established based on Bayesian network. The diagnostic method considering experts’ knowledge and maintenance experience is presented as well as the network inference procedure. The proposed method can improve the diagnostic efficiency and accuracy.