基于键合图和动态贝叶斯网络的逆变器系统可靠性研究  

Research on Reliability of Inverter System Based on Bond Graph and Dynamic Bayesian Network

作　　者：程硕 帕孜来·马合木提[1] CHENG Shuo;MAHEMUTI Pazilai(College of Electrical Engineering,Xinjiang Univ.,Urumqi 830047,China)

机构地区：[1]新疆大学电气工程学院,乌鲁木齐830047

出　　处：《三峡大学学报（自然科学版）》2022年第6期101-107,共7页Journal of China Three Gorges University：Natural Sciences

基　　金：国家自然科学基金(61963034)。

摘　　要：针对逆变器系统级的可靠性问题,提出一种键合图理论结合动态贝叶斯网络的可靠性分析方法,以中点钳位型三电平逆变器(neutral point clamped, NPC)为研究对象开展可靠性分析研究.首先,基于工作原理建立系统的键合图模型.然后,考虑到键合图理论因果关系与贝叶斯网络因果关系的相似性,搭建初始的静态贝叶斯网络,通过引入马尔可夫过程,得到系统关键元件的状态转移概率,建立系统的动态贝叶斯网络模型.最后,用Genie软件对提出的可靠性分析方法开展动态仿真.研究结果表明:所采用的方法不仅能反映逆变器系统可靠性的动态变化情况,还能实现对系统元件在不同状态下的故障定位,对逆变器系统的可靠性评估与健康管理具有积极意义.Aiming at the reliability of inverter system, a analysis method of the reliability based on bond graph theory and dynamic Bayesian network is proposed in this paper. The reliability analysis is carried out with the research background of neutral point clamped three-level inverter(Neutral Point Clamped, NPC). Firstly, the bond graph model of the system is established based on the working principle. Then, the initial static Bayesian network is established under considering the similarity between the causality of bond graph theory and Bayesian network. Furthermore, the state transition probability of key components in the system is obtained by introducing the Markov process, and the model of dynamic Bayesian network is established. Finally, the proposed reliability analysis method is dynamically simulated with Genie software. The results show that the method can not only reflect the dynamic changes of the reliability of inverter system, but also realize the fault location of system components in different states, which is of positive significance to the reliability evaluation and health management of inverter system.

关 键 词：键合图理论 NPC逆变器 动态贝叶斯网络 系统可靠性 动态分析 

分 类 号：TM933[电气工程—电力电子与电力传动]