Iterative learning control in prosumer-based DC microgrids

As part of the energy transformation to 100 % renewable energies, decentralized energy supply, the generation of electrical energy close to the consumer is becoming increasingly popular. This includes islanded networks, that are, e.g., prosumer-based, i.e., producing and consuming simultaneously at each grid node. The increasing use of DC-based resources and loads is a relevant solution for electrification of rural areas in the global south and beyond. DC is preferred because of lower losses at longer distances in the transmission compared to AC and reduced inversion losses. Load profiles in energy networks often have periodic components. These can be learned a day ahead by an appropriate control, so that instantaneous control power of a system can be saved. Iterative learning control (ILC) is suitable for this purpose. In this paper, ILC is applied in prosumer-based DC microgrids to investigate the above-mentioned requirement of balancing periodic components. For this purpose, a model for a droop-controlled prosumer-based DC microgrid is proposed and described with suitable equations and system parameters as one level, and as a second control level the ILC is connected to it. The overall model is simulated and its performance validated with respect to the control objectives of bounded voltage deviation, power sharing and reduced low-level control energy. Interestingly we observe that the load sharing properties of the low-level control can enable learning without full information. The implicit communication contained in the low-level is sufficient. This shows that our design can operate in a decentralized fashion without additional communication infrastructure.