Cooperative control of multi-agent systems: Where additional communication links improve the transient behaviour

Abstract

The paper presents an interesting structure of cooperative control in which every additional communication link improves the transient behaviour of the overall system. The performance output is the weighted sum of the agent outputs and the transient behaviour is evaluated by a quadratic performance index. The agents may send their control error to a neighbouring agent, which uses this information to modify its own control error, which is the input to the local controller. The main result shows that in a path communication graph with the disturbed agent as the root vertex any additional communication link that prolongates the path improves the overall system performance. This result is demonstrated by its application to a simplified model of an electrical power system.