Multi-Agent System Based Distributed Detection And Fault Localization

Multi-agent System has assumed increasing significance in the field of artificial intelligence.In order to perform different complex works in the multi-agent system with the continuous time, it needs the real-time distributed change detection of multi-agent systems and the fault agent localization.It contains three parts in the thesis. The first part investigates the consensus based distributed change detection of the multi-agent systems. The second part extends decentralized algorithm to estimate the eigenvalues of the Laplacian matrix that encodes the network topology of a multi-agent system. The third part proposes an adaptive fault localization algorithm in the continuous time.First, we investigate consensus detection of the distributed multi-agent systems. This is a typical problem which consists of detection of the distributed multi-agent systems in the real time and the estimate of the mutation time. In the paper, the algorithm is based on a combination of recursively generated local statistics and a global consensus strategy.It does not require any fusion center.Second, we investigate the distributed multi-agent system topology detection. It achieves the connection of the agent nodes in the multi-agent system by estimating topology. The basic idea is to provide a local interaction rule among agents so that their state trajectory is a linear combination of sinusoids oscillating only at frequencies function of the eigenvalues of the Laplacian matrix. The topological structure can be estimated in the continuous time based the characteristic of the basic topology. What’s more, the fault of the multi-agent system can be diagnosed.Third, we propose the fault localization in the multi-agent systems.When the agent is broken by the energy depletion or the circuit fault, the agent malfunctions. The fault can not maintain the consensus movement as the neighbor agent in the multi-agent network. In order to locate the fault agent in the continuous time, the paper proposes an adaptive localization algorithm which permits a mobile localization agent to estimate the location with the measured distance from the fault target in the continuous time. In addition, the paper also discusses PE property in the localization algorithm. It can achieve the good effects.All of above demonstrated that, when the multi-agent system state change occurs, the proposed strategy is capable of monitoring uncertainty real time multi-agent system. Through the multi-agent system consists of mutation detection and the corresponding fault agent detection, it can achieve high-quality multi-agent control system.