The study presents an integrated adaptive data-based system that manages load demands and schedules outages in contemporary distribution systems. The proposed framework enables complete understanding of its operation because it combines three elements that help to predict future events, evaluate system performance, and make decisions to reduce operational demands. The system uses three supervised learning modules, which include load forecasting based on different scenarios and SVM-based sensitivity classification with multiple features and service probability assessment through multi-level logistic regression to create results that run through a neighborhood optimization engine. The system achieves fairness by reallocating service interruptions to low-sensitivity loads while maintaining uninterrupted service to essential facilities. The IEEE 33-bus system simulations achieved operational improvements through 41% reductions in SAIFI and 43% reductions in SAIDI, which resulted in better system performance and more balanced power outages. The framework establishes three operational attributes that promote transparent and fair decision-making to achieve resilient outage management through socially acceptable methods. The framework operates within SCADA/ADMS systems because its lightweight design and interpretability enable SCADA/ADMS systems to implement intelligent distribution network operations. • An adaptive ML architecture enables fair, resilient, real‑time load shedding in modern smart distribution networks. • A transparent layer uses regression, adaptive SVM, and logistic models for forecasting and interpretable decisions. • Scenario‑driven scheduling protects critical loads and redistributes outages across low‑sensitivity areas (IEEE 1366). • IEEE 33‑bus validation shows 41% SAIFI and 43% SAIDI reduction with zero curtailment of critical consumers. • The framework enhances equity and resilience and integrates with SCADA, PMU, and ADMS for real deployment.
Sanaz Ghanbari (Mon,) studied this question.