基于MAS的实时诊断技术在电源系统中的应用研究  

Applied Research of Real Time Fault Diagnosis Technology for Power System Based on MAS

作　　者：安若铭[1,2] 安伟光 

机构地区：[1]哈尔滨工业大学航天学院,哈尔滨150001 [2]哈尔滨工程大学建筑工程学院,哈尔滨150001

出　　处：《宇航学报》2008年第6期2068-2072,共5页Journal of Astronautics

摘　　要：为了提高卫星这类大规模复杂系统的诊断速度和精度,提出了一种基于MAS的卫星电源系统分布式诊断框架。框架的核心部分是两类诊断Agent:一类是自适应全局Agent,负责全局诊断,采用多信号模型的诊断方法;另一类是子系统驻留Agent,负责局部诊断,采用参数估计的诊断策略,并引入Box-Jenkins的建模方法。通过通讯Agent和电源专家Agent实现多Agent间协调工作。在电源系统中的应用表明,两种诊断Agent间优势互补,提高了诊断效率,并能准确地给出诊断结果。For improving diagnosis speed and accuracy of large scale and complex system like satellite, a distributed diagnosis frame of satellite power system is put forward based on MAS. The most important parts of the frame arc two types of diagnosis agent. One is adaptive global agent and responsible for global diagnosis, and muhi signal model is adopted to diagnosis; another is subsystem resident agent and responsible for local diagnosis, and parameter estimation diagnosis strategy is adopted. Box-Jenkins modeling method is introduced. Through communication agent and power expert agent, coordination operation between agents is realized. The results of applied to power system show that goodness of two types of diagnosis agents are complementary, the diagnosis efficiency is improved and diagnosis outcomes are accurate.

关 键 词：多Agent系统(MAS) 参数估计 B-J模型 多信号模型 故障诊断 电源系统 

分 类 号：V474[航空宇航科学与技术—飞行器设计] TP391[自动化与计算机技术—计算机应用技术]