| Multi-agent systems have found wide applications in the fields such as intelligent transportation,smart grid and distributed sensor networks and so on.The cooperative control of multi-agent systems aims at designing distributed control protocol,with which all the agents can share a common value eventually and solve complex tasks that single agent cannot operate.However,privacy leakage is inevitable in the scenario of interacting with their neighbors.Due to this concern,the privacy protection issue of multi-agent systems has attracted extensive attention.Based on the state-of-the-art results on the cooperative control of multi-agent systems,this thesis investigates the coordination control and privacy protection for multi-agent systems with antagonistic information.The main contents are listed below:First,the cooperative control and privacy protection of first-order multi-agent systems with antagonistic information are studied over an undirected connected signed graph.For the structurally balanced communication topology,some sufficient conditions for bipartite consensus in the sense of mean square and almost sure are derived,respectively.The eventual convergence value is obtained,an ε-differential privacy algorithm is proposed,and the optimal noise is designed.Furthermore,the trade-off between system performance and the degree of privacy protection is discussed.Second,based on the above results,we further investigate the cooperative control and privacy protection of first-order multi-agent systems with antagonistic information where the topology is characterized by strongly connected signed graph.In this instance,the Laplacian matrix corresponding to the directed topology no longer has symmetry.Two cases involving structurally balanced graph and structurally unbalanced graph are considered.As for the structurally balanced case,an ε-differential privacy algorithm is proposed,upon which some criteria guaranteeing almost sure bipartite consensus are given.The tradeoff between system performance and the privacy guarantee is elaborated,and the optimal noise is also devised.Then the proposed privacy preserving scheme is further applied to the scenario of structurally unbalanced graph,where a criterion with respect to almost sure stability of the considered system is derived as well as the privacy preserving condition.Finally,the cooperative control and privacy protection results are extended to second-order multi-agent systems with quasi-strongly connected graph.The Laplacian noise is injected for both position and velocity states.As for structurally balanced case,some necessary and sufficient conditions for almost sure bipartite consensus are given,different from the first-order multi-agent systems,in addition to the condition that the topology contains a directed spanning tree,the necessary and sufficient conditions are also related to system parameters and other factors.Upon which an ε-differential privacy algorithm is developed.Then the tradeoff between system performance and the degree of privacy protection is discussed,and the optimal noise is also elaborated.As for structurally unbalanced case,some criteria for almost sure stability or interval bipartite consensus are induced. |
