Risk-Aware Multi-Timescale Scheduling and Distributed Event-Triggered Control for Overload Mitigation in PV-Rich Distribution Networks

Abstract

: High-penetration distributed photovoltaic (PV) integration improves renewable utilization in distribution networks, but abnormal PV ramps may also be transformed into feeder active-power overloads when coupled with load variation and feeder power-flow baselines. To address this issue, this paper proposes a risk-aware multi-timescale scheduling and distributed event-triggered control framework for PV-rich distribution networks. The proposed framework links deterministic forecasts, probabilistic risk information, and real-time measurements with day-ahead scheduling, intraday rolling correction, and real-time distributed control, respectively. In the day-ahead stage, multiple flexible resources are economically dispatched based on point forecasts to establish the operating baseline and mitigate foreseeable overloads. To quantify residual PV uncertainty, a risk-boundary increment is defined as the difference between the probabilistic and point-forecast feeder load rate, with the former derived from a probability prediction method based on quantile regression forest (QRF). The resulting risk-boundary increment is then converted into the adjustable reserve margin for intraday rolling correction, enabling risk-adaptive flexibility allocation. In the real-time stage, a distributed event-triggered controller is activated by measured feeder overloads to suppress residual minute-scale violations while reducing unnecessary communication. Case studies on the IEEE 33-bus system show that intraday correction reduces the feeder overload rate from 47.9% to 1.04%, while real-time control eliminates the remaining overloads. Event-triggered communication retains the same control performance with only 2.2% of the communication required by periodic control. A preliminary test on an actual regional distribution network further supports the transferability of the proposed framework.