Research On Fault Diagnosis Technology For Multi-agent Systems With Switching Topologies

With tremendous military value and broad prospects for civilian applications,multi-agent systems have become the focus of research around the world.Multi-agent systems are mainly applied in the fields of mobile robot formation and aircraft formation.Affected by the operating environment,aircraft and robots are prone to failure.Therefore,how to design an effective fault diagnosis strategy to ensure the safe operation of multi-agent systems under fault conditions is a very challenging and significant issue.In this paper,the fault diagnosis method is studied for the multi-agent systems under the premise of system topology switching.First of all,the background and research significance of the subject are introduced.The status quo of multi-agent system development is elaborated.The research status of switching topological control technology and multi-agent fault diagnosis technology are also summarized.Secondly,considering the characteristics of switching topologies,the fault estimation strategy for multi-agent systems under undirected topological structure has been studied based on model-dependent average dwell time.Unknown interference and actuator failure are assumed in the multi-agent systems.The distributed observer is firstly designed based on the information of the adjacency agent.The relationship and influence between each agent and observer are analyzed.Then the global dynamic estimation error equation is constructed.On this basis,combined with the robustness theory,the observer parameters are designed for relaxed switching frequency conditions to ensure the system status monitoring in the presence of external disturbances and actuator faults.Thirdly,according to the extensive application of multi-agent technology,the fault detection method of the nonlinear multi-agent systems with switching topologies is studied based on the interval observers.The interval observers are designed for the system under the fixed topology and switching topology respectively to study the fault detection strategy.Taking the assumptions and limitations of the interval observer into account,an improved interval observer is designed by using a linear transformation method to improve the applicability of the fault diagnosis method.The existence of the interval observer and the method of parameter design are given based on the theoretical proof of the fault diagnosis method.Finally,further research on fault diagnosis methods for multi-agent systems with switching topologies is studied.Considering that the system operating environment may not keep the topology structure connective,an improved fault diagnosis strategy based on the average system theory is proposed.Compared with the previous research,the proposed strategy can achieve fault estimation of multi-agent systems under the joint connected topologies.The proposed method has been rigorously proved theoretically.The solution to the linear matrix inequalities for observer parameters is also given.