Study On Consensus Of Heterogeneous Multi-agent Base On Model Reference

In recent years,in view of its many applications,such as cooperative control of unmanned aerial vehicles,communication between sensor networks and cooperation of intelligent robots,multi-agent consistency has gradually become an active research field,in which system consensus has become the focus of research on the premise of ensuring the stability of the system.The research of multi-agent is divided into isomorphic multi-agent and heterogeneous multi-agent.Compared with heterogeneous multi-agent systems,the research on isomorphic multi-agent systems has been quite fruitful,but considering the actual application,the structure of the agents in the system is often different,so considering the situation of heterogeneous multi-agent can better meet the needs of practical application.In this paper,for heterogeneous linear multi-agent systems with unknown parameters,a fixed output average consistency protocol is adopted in undirected or balanced directed networks,so that the output of each agent reaches the average of their initial output.Each agent in the network is modeled as an unknown linear system with different order and correlation of 1 or 2,and its state is updated according to its own output and the output of neighboring nodes.Based on the model reference control method,the corresponding models are defined for the agents with different correlations,and then the consistency protocol is proposed to converge the output of each agent to the output of its reference model,that is,to achieve the average consistency of the fixed output.Finally,an illustrative example is used to demonstrate the effectiveness of the proposed protocol,and two groups of parameters are selected to observe the impact on the convergence rate.Compared with the existing work,the upper limit of the unknown parameters of the system is unknown.The correlation degree of each agent is different.The high-frequency benefit of the agent with correlation degree 2 is unknown.In addition,because the expected value of output is the average value of each agent's initial output,the restriction on each agent's output is relatively relaxed.