为解决复杂系统的多故障容错问题,首先将多智能体系统(MAS)的基本概念进行了扩充,定义了容错多智能体系统(FATMAS)的相关概念,并在此基础上提出了在FATMAS中实现多故障容错的各类函数定义、规则与流程。然后按照一定的流程调用不同函数就可对系统中的多故障进行检测,并可通过复制非关键智能体的任务使系统从多故障中恢复运行。F-16飞机多故障诊断与自修复的仿真结果说明,该方法能够通过清晰的智能体间消息传递机制有效地减少系统中复制智能体的数目,而且比较接近于实际系统中的故障处理过程,可用于复杂系统的多故障诊断与容错。
To solve the problem of multi-fault diagnosis of complex systems,the basic concept of Multi-Agent Systems(MAS) is expanded to establish the concept of Fault-Tolerant Multi-Agent System(FATMAS).Based on which the definitions of various functions,rules and flows for realizing multi-fault tolerance in FATMAS are given.Then the multiple faults can be detected by calling different functions of the system according to some definite flows,and the system can recover from the faults by copying the tasks of non-critical agents.The results of simulation in the F-16 aircraft multi-fault diagnosis and self-repairing illustrated that this method can effectively reduce the number of replicated agents through clear inter-agent message passing mechanism,and it is relatively close to the process of dealing with failures in actual system,therefore it can be used in complex system for multi-fault diagnosis and tolerance.