摘要
混杂系统故障诊断的关键在于系统模型的建立及状态的估计。利用粒子滤波在混杂系统状态估计上的优势和键合图易于系统建模的特点,提出了一种基于粒子滤波和键合图理论的混杂系统故障诊断方法。首先把系统用键合图语言描述,进而得出系统的状态方程。然后利用粒子滤波算法,同时估计系统的离散和连续状态,当残差超过阈值时,认为系统发生故障。最后在三容水箱传感器故障诊断中验证了该方法的有效性。
Model establishment and state estimation are the key points of fault diagnosis for hybrid systems. By using the advantages of particle filtering in state estimation for hybrid system,and the feature of bond graph for establishing system model easily,the fault diagnosis method based on particle filtering and bond graph theory for hybrid system is proposed. Firstly,the system is described using bond graph language,and then come to the state equation of the system. By adopting particle filtering algorithm,the discrete and continuous states of the system are estimated simultaneously; when the residual value exceeds the threshold,then system failure will be considered. The effectiveness of this method is verified in fault diagnosis of the sensors in three-tank system.
出处
《自动化仪表》
CAS
2015年第12期12-15,共4页
Process Automation Instrumentation
基金
国家自然科学基金资助项目(编号:61364010)
关键词
混杂系统
电磁阀
传感器
粒子滤波
键合图
状态估计
故障诊断
Hybrid system
Electromagnetic valve
Sensor
Particle filtering
Bond graph
State estimation
Fault diagnosis
