In this study, we propose a data-driven control method that compensates for the effects of nonlinear terms with a simple configuration. Although the control target that can be handled is limited, the proposed method has an advantage in that the conventional FRIT can be extended without using special conversion. The effectiveness of the proposed method was verified by numerical simulations for two examples: a simple inverted pendulum and a shaking-table that is close to a real problem. In the example of a shaking-table, we confirmed that the effect of nonlinear friction can be suppressed appropriately. Therefore, we believe that the proposed method can be one of the practical methods.
In graph exploration via multi-agent systems, individual agents make decision depending on surroundings environment and state of other agents and achieve the aim of the agent and system. This paper evaluates the relation between the role exchange between agents for deadlock resolution and exploration efficiency. If a communication state is bad, the role exchange may cause disappearance and replication of the role. This paper formulate outbreak condition of the deadlock and let the condition of the disappearance and replication of the consensus condition of role exchange. This paper proposes a communication protocol which used the deadlock condition. The first result is to have enabled the consensus of role exchange not to depend on the communication state. The second result is to show that the disappearance and replication of the role may cause a agglutination state of the exploration system.