贝叶斯网络和模糊评判结合的滚动轴承故障诊断
2018-12-21马德仲任锁刘凯辛李明周真
马德仲 任锁 刘凯辛 李明 周真
摘要:针对大型复杂系统在诊断的过程中,由于现有方法主要通过一系列方法来提高诊断的效果,而缺乏考虑诊断过程中的检测难度、检测速度和检测经济性等因素。提出了贝叶斯网络诊断与多因素模糊综合评判相结合进行故障诊断的方法,在诊断的过程中不仅考虑故障概率,而且结合检测方法难易程度、检测速度、检测的准确性和经济性等方面,得到诊断的优化方法。通过对齿轮箱滚动轴承故障进行诊断的实例,可以明显看出该方法在综合诊断过程中的优势。研究成果可以作为对大型复杂系统进行故障诊断的优化方法,从而科学指导维修方案。
关键词:
贝叶斯网络;模糊综合评判;诊断优化
DOI:10.15938/j.jhust.2018.05.019
中图分类号: X9287
文献标志码: A
文章编号: 1007-2683(2018)05-0113-06
Fault Diagnosis of Rolling Bearing Based on Bayesian Network and Fuzzy Evaluation
MA Dezhong,REN Suo,LIU Kaixin,LI Ming,ZHOU Zhen
(School of Measurementcontrol Technology and Communications Engineering, Harbin University of Science and Technology,
The Higher Educational Key Laboratory for Measuring and Control Technology and Instrumentations of Heilongjiang Province, Harbin 150080, China)
Abstract:In the process of diagnosing large and complex system, the existing methods mainly improve the diagnosis effect by a series of methods, but lack the consideration of the difficulty of detection, the detection speed and the detection economyIn this paper, Bayesian network diagnosis and multifactor fuzzy comprehensive evaluation are combined to diagnose the fault In the process of diagnosis, not only the probability of failure is considered, but also the difficulty of detection, the speed of detection, the accuracy and economy of detection, and the optimization method is obtainedThe advantages of this method in the comprehensive diagnosis process can be clearly seen by the example of diagnosing the fault of gearbox bearingThe research results can be used as an optimization method for fault diagnosis of large and complex systems, so as to guide the maintenance plan scientifically
Keywords:bayesian network; fuzzy comprehensive evaluation; diagnostic optimization
0引言
隨着生产力的快速发展,越来越多的大型设备投入使用,运行系统由简单变得越来越复杂化,同时系统故障也越来越多样化和复杂化,因此需要对复杂系统的故障和危险进行有效的诊断,减少故障时间[1]。近年来,基于图论的许多诊断方法得到了学术界和工业界的快速发展,如键合图、信号流图、故障树和有向图等[2-5]。贝叶斯网络是由一系列变量和有向边组成的有向无环图,对于处理不确定性的概率知识表达和推理具有巨大作用[6],并且已被广泛应用于各种领域的故障诊断。韩璞等[7]提出贝叶斯网络和主成分分析法相结合,通过对主成分分析方法提取故障特征的讨论,建立贝叶斯网络故障诊断模型,实例表明,该方法模型简洁、易于推理,提高了汽轮机故障诊断的效率。……
