Monte Carlo-based uncertainty framework for QCNN-guided hybrid optimization in distribution system load shedding

This study proposes a priority-aware load-shedding framework for radial distribution systems with photovoltaic generation, wind generation, and electric-vehicle charging demand under uncertain operating conditions. The objective is to determine the required amount of load curtailment while reducing unnecessary interruption of important loads and keeping feeder voltages within the acceptable ± 5% range. Monte Carlo simulation is used to represent variations in load demand, photovoltaic output, wind generation, and EV charging demand. For each generated scenario, a Quantum Convolutional Neural Network (QCNN) evaluates the priority of each load bus using normalized load magnitude, voltage sensitivity, renewable-source proximity, and customer type. The obtained priority scores are then included in a normalized multi-objective optimization model solved using a hybrid American Zebra Optimization Algorithm–Red Panda Optimization Algorithm (AZOA–RPOA). The objective function considers economic curtailment cost, target-shedding accuracy, priority preservation, voltage-limit violation, and real power loss in a dimensionally consistent manner. The proposed framework is evaluated on a modified IEEE 33-bus radial distribution system using 100 Monte Carlo scenarios and compared with a rule-based baseline, ablation variants, AHP–PSO, and a DRL-based benchmark. Compared with the baseline, the proposed QCNN–AZOA–RPOA method reduces the mean economic cost by 37.71% and the mean real power loss by 42.14%, while maintaining all scenario voltages within 0.95–1.05 p.u. It also provides the lowest priority-weighted shedding among the compared methods, showing better protection of high-priority loads. Overall, the results indicate that the proposed framework offers a balanced load-shedding solution by combining economic performance, voltage security, and priority-aware curtailment under RES and EV uncertainty.