A Machine Learning-Driven H-DERMS Framework to Manage and Control Multiple DERs

The proliferation of distributed energy resources (DERs) has introduced significant challenges in power systems control and management. This paper presents a hierarchical distributed energy resource management system (H-DERMS) framework that integrates machine learning (ML) to effectively tackle these challenges. The proposed framework comprises utility and aggregator DERMS layers, effectively managing directly connected and aggregated DERs. The management problem is formulated as an optimal power flow (OPF), which is simplified using a Sequential Adjustment Method and Random Forest Regression to address the large-scale, nonlinear, and non-convex nature of the OPF problem to make it suitable for real-world applications. This simplification not only enhances computational efficiency but also yields improved solution quality. Tested on the IEEE 33-node and 123-node distribution networks, the proposed approach demonstrates robust performance across diverse case studies, showcasing its effectiveness in handling the complexities of DER management and underscoring its potential for practical deployment in modern power systems.