Consensus And Containment Control Problems Of Multi-agent System Based On Predictive Control

Recently, the cooperative control of multi-agent system becomes one of hot issues in thefield of control theory. The research in biology shows that natural bio-groups have theadvanced predictive intelligence and every individual has the ability to adopt the neighbor’spast trajectory and the current available information to determine its future motion. This thesismainly studies the cooperative control of multi-agent system based on predictive control. Themain contents are as follows.1. The predictive consensus of single-integrator multi-agent system with fixed topologyis investigated. A distributed control protocol with predictive term is presented, and analgebraic-type necessary and sufficient condition is obtained under which consensus isachieved. Compared with traditional algorithm, the results of simulation illustrate that theproposed algorithm has faster convergence speed and the sampling period is extended.2. The predictive containment control of single-integrator multi-agent system isconsidered. A predictive control-based distributed control algorithm for multi-agent systemwith multiple static leaders is given and a necessary and sufficient condition for containmentis acquired for the system with fixed topology. For the switching topology case, the analysisshows that the predictive term can significantly improve the convergence speed of the systemas long as the union of the interaction graphs has a spanning tree frequently enough.3. The predictive containment control for a class of second order multi-agent system withdynamic leaders is investigated. A distributed control protocol with predictive control termfor second order multi-agent system is proposed. The impact of each parameter on theperformance of the system to solve the containment control problem is concluded and therange of parameters selection is obtained by regularly changed the corresponding controlparameters. Compared with the system without prediction term, the results of simulationillustrate that all the followers are able to quickly converge to the convex hull spanned by theleaders, and then move with the leaders with the same velocity.