| With the rapid development of the computer,communication and wireless sensor technology,distributed multiagent systems has attracted considerable attention and become a hot topic in field of artificial intelligence and control engineering.Multiagent systems(MASs)consist of lots of mobile agents which have simple communication,decision-making and executive ability.Agents communicate and cooperate with neighbors,which MASs can enhance the ability of solve complex task and achieve destination task which single mobile agent can't achieve.The first priority of distributed cooperation control for multiagent systems is to reach consensus.Olfati-Saber proposed a distributed discrete-time consensus protocol which requires each mobile agent communicates with all neighbors during evolving.However,communication redundancy in the evolution of large-scale systems restricts consensus and convergence speed.Therefore,it is necessary to design a reasonable and effective neighbor selection strategy for consensus protocol to reduce unnecessary communication.Preserving connectivity of communication topology is a necessary condition for MAS s to reach consensus.Thus,it is particularly important to study the algorithm of maintaining connectivity for the MASs with time-varying topology.The main contents and results are as follows.The rendezvous problem of discrete-time multiagent systems under bi-directional switching topology is investigated.A framework of distributed consensus protocol,based on neighbor selection strategy,where each agent evolves by only reference to the state of some neighbors,is first proposed.Next,the nearest neighbor selection strategy(NNS)is proposed and the NNS is optimized using distributed algorithm.Then,the least common neighbor selection strategy(LCNS)is proposed.The new distributed consensus protocol enhances the MASs' consensus and embodies the agent's intelligence and the ability to assess the situation.The NNS is that each agent actively selects a neighbor who has the minimum difference of state in each sector by communication sector strategy.In the LCNS,each agent actively selects the neighbor who has the least number of common neighbors with the agent.Then,the stability of the consensus protocol with the NNS and the LCNS is analyzed with the common Lyapunov function and numerous simulations.The range of parameters affecting the convergence of consensus is discussed.Finally,the necessary condition for enhancing connectivity is obtained through numeric analysis with massive mobile agents.Compared with classical consensus protocol,the new distributed consensus protocol enhances the MASs' consensus.Simultaneously the requirements for data storage and computational load are all significantly reduced.Further,the energy consumption is also shown to have been reduced due to the maintenance of connectivity,without extra control algorithms.The distributed consensus protocol for large-scale systems,based on connectivity preservation algorithm,where each agent evolves by only reference to the state of the virtual neighbors is proposed in following chapter.Virtual neighbors were constructed by equally communication sector strategy,on this basis,combining with connectivity preservation algorithm.Therefore,the agents only can move in motion area to maintain initial communication links for connectivity preservation and reach consensus.Theoretical analysis and numerous simulations were combined to verify the validity of the new protocol.Compared with classical consensus protocol,the results of simulations pointed out that the number of clusters decreases,convergence speed of the MASs could be increased. |
