Multiagent Control in Modular Motor Drives by Means of Deterministic Consensus

A modular motor drive (MMD) is composed of multiple identical pole drive units (PDUs), which consist of a concentrated stator winding fed by a dedicated power electronic converter module. To exploit the full flexibility and reliability of these MMDs, they are equipped with multiple controllers, operating as synchronized peers, and each driving only a limited set of PDUs. An MMD can hence be considered as a multiagent system, in which all the agents cooperate to deliver a certain torque demand. The goal of this work is to distribute this torque demand among the agents of a modular drive in a completely decentralized way, in order to comply with the flexibility and reliability of the hardware, while still taking into account the present condition of each agent. A deterministic consensus algorithm is used for this purpose, in combination with decentralized proportional integral (PI) current controllers. Communication is only allowed between neighboring agents, to limit the communication load. It is shown in this article that the total torque demand is delivered, under healthy conditions as well as during agent malfunctions and with nonidentical agents. The concept is demonstrated on a modular axial-flux permanent magnet synchronous machine with 15 PDUs that are assigned to five agents.