摘要
针对双组元推进系统中环境复杂且部分关键信息无法通过传感器获取,提出基于部分可观时间Petri网故障诊断性的贝叶斯估计方法。研究了基于部分可观时间Petri网的双组元推进系统故障诊断性的贝叶斯估计问题。系统过程的变迁分为可观和不可观,结合变迁触发关系和变迁时间信息,建立Petri网模型的状态类图。遍历所有满足可观测变迁触发时间和序列信息的路径,对于诊断结果为可能故障系统,建立故障变迁对应的贝叶斯Petri网模型,将不可观变迁设置为贝叶斯变迁,根据可观变迁触发状态估计不可观变迁触发概率,进一步判断系统故障状态。最后,建立了基于部分可观Petri网的整体推进系统模块,通过仿真实验验证了算法的有效性。
To solve the problems that the environment of bipropellant propulsion system is complex and some key information can’t be obtained through the sensors,a method of Bayesian estimation of fault diagnosis based on POTPN was proposed.In this study,we investigated the Bayesian estimation of fault diagnosis based on partially observable time Petri net(POTPN)for the integral bipropellant propulsion system.The transitions were divided into observable transitions and unobservable transitions.The state class diagram of the Petri net model was established by combining the transition trigering relationship and the transition time information.Searching all paths that are valid by the trigering time of the observable transitions and the sequence information.For the system diagnosed as a possible fault,a Bayesian Petri Net(BPN)corresponding to the fault transition was established,setting the unobservable transition as the Bayesian transition,and estimating the unobservable transition trigering probability according to the trigering state of observable transition to further judge the fault status of the system.Finally,the module of integral bipropellant propulsion system based on the POTPN was built,verifying the validity of the algorithm with the data in simulation experiment.
作者
张信哲
张治国
丁晓彬
刘久富
杨忠
王志胜
ZHANG Xinzhe;ZHANG Zhiguo;DING xiaobin;LIU Jiufu;YANG zhong;WANG zhisheng(School of Automation,Nanjing University of Aeronautics and Astronautics,Nanjing 210016,China)
出处
《应用科技》
CAS
2020年第1期61-67,共7页
Applied Science and Technology
基金
国家自然科学基金项目(61473144).
关键词
航天推进系统
故障诊断
部分可观
时间PETRI网
贝叶斯网络
状态类图
线性规划
贝叶斯变迁
aerospace propulsion system
fault diagnosis
partially observed
time Petri nets
Bayesian nets
state class diagram
linear programming
Bayesian transition
