贝叶斯网络在空间机械臂故障诊断中的应用研究  

Research on Application of Bayesian Network in Fault Diagnosis of Space Manipulator

作　　者：毛保磊[1] 慕德俊[1] 张慧翔[1] 肖旭[1] 刘建歌[1] 

机构地区：[1]西北工业大学自动化学院,陕西西安710072

出　　处：《计算机技术与发展》2013年第8期27-30,共4页Computer Technology and Development

基　　金：国家自然科学基金资助项目(60803158);西北工业大学研究生创业种子基金(Z2012120)

摘　　要：空间机械臂在空间设施中广泛应用,如何准确快速判断其运行状态成为需要解决的重要问题。文中首先介绍了机械臂在国内外空间设施的应用背景,然后采用基于故障树与贝叶斯网络的故障诊断方法,包括:基于系统组成结构的故障树建模、贝叶斯网络转化、团树传播算法和最大后验估计(MAP)推理结果分析。最后利用单点故障实验和多点故障实验对所提方法进行了验证,结果表明将故障树分析和贝叶斯网络结合应用到空间机械臂故障诊断是有效的、可行的。Space manipulator has been widely used in outer space facilities, how to judge its state fast and correctly is the most important and difficult problem. It introduces the applied background of manipulator in outer space facilities both in China and abroad briefly. Then it illustrates the fault diagnosis procedure combining Bayesian network with fault tree, which includes constructing fault tree model based on system structure,Bayesian network transformation, clique tree propagation and MAP inference result analysis. Finally, verify it with single-point and multi-point fault experiments and demonstrate it is effective and feasible by combining fault tree and Bayesian network in manipulator fault diagnosis.

关 键 词：空间机械臂 故障诊断 故障树 贝叶斯网络 团树 最大后验估计 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]