Deep multi‐agent Reinforcement Learning for cost‐efficient distributed load frequency control

The rise of microgrid-based architectures is modifying significantly the energy control landscape in distribution systems, making distributed mechanisms necessary to ensure reliable power system operations. In this article, use Reinforcement Learning techniques proposed implement load frequency (LFC) without requiring a central authority. To end, detailed model dynamic behaviour formulated by representing individual generator dynamics, rate and network constraints, renewable-based generation, realistic realisations. LFC problem recast as Markov Decision Process, Multi-Agent Deep Deterministic Policy Gradient algorithm used approximate optimal solution all layers, that is, primary, secondary tertiary. framework operates through centralised learning implementation. particular, there no information interchange between generating units during operation. Thus, communication infrastructure privacy them respected. validated numerical results it shown can be cost-efficient manner.