Probabilistic Estimation of PV Hosting Capacity in DER-Rich Feeders Using Smart Meter Data

The growing penetration of photovoltaic (PV) systems in distribution networks (DNs) highlights the significance of PV hosting capacity (HC) estimation for system planning and operation. Compared to traditional model-based methods, data-driven HC estimation offers superior computational efficiency and scalability, gaining widespread attention. This paper aims to address the limitations of previous data-driven approaches for residential PV HC estimation in DER-rich feeders by proposing a probabilistic HC estimation framework. The framework first determines transformer pairings using available smart meter (SM) data and limited network information. Next, an optimization-based voltage sensitivity estimation model accurately estimates voltage sensitivity at customer nodes in the target low-voltage secondary network (SNet). Finally, a Gaussian mixture density network characterizes the head bus voltage of each SNet and derives the distribution of the HC at customer nodes. Numerical results and method comparisons on the EPRI Ck5 circuit validate the effectiveness of the proposed framework.