Coordinated dispatch of virtual power plants in distribution systems under ramping constraints and uncertainties

Abstract

This paper proposes a coordinated dispatch framework for distribution systems with virtual power plants (VPPs) that addresses both substation ramping constraints and load-ramping uncertainty. A data-driven quantile-regression temporal convolutional network (QR-TCN) is developed to produce multi-quantile probabilistic forecasts of both the load and its ramping boundaries. Using the resulting prediction intervals, a second-order cone programming (SOCP) dispatch model is formulated to jointly minimize upstream substation operating costs, VPP regulation costs, and network losses. Case studies on the IEEE 33-bus system demonstrate that the proposed framework ensures system economic, flexibility, and security adequacy across diverse operational boundaries and system constraints.