Abstract <p>The extensive integration of renewable energy sources has heightened the unpredictability, uncertainty, and operational complexity of contemporary power systems, rendering traditional deterministic optimal power flow (OPF) methods inadequate for secure and affordable operation. This research presents a dynamic stochastic multi-objective optimal power flow (DS-MOOPF) framework employing a hybrid global-best-guided Artificial Bee Colony and Non-dominated Sorting Genetic Algorithm-II (gbestABC–NSGA-II) for renewable-integrated power systems. Wind speed, solar irradiation, and load demand uncertainties are modeled with Zhao’s Point Estimation Method (Zhao’s PEM) in conjunction with Nataf Transformation (NT). The framework concurrently optimizes operational costs, emissions, active power loss, voltage magnitude deviation, voltage stability, loadability, small-signal stability, and transient stability targets. The primary methodological innovation consists of integrating gbest-guided ABC search with NSGA-II elitist sorting, a reinforced constraint-dominance principle (SCDP) for effective constraint management, and a decomposition-based archiving approach (DAA) for sustaining a bounded and uniformly distributed Pareto archive. An integrated Interval Type-2 Fuzzy AHP-TOPSIS ( IT2FAHP-TOPSIS) module is employed for the selection of the optimal compromise solution by amalgamating stakeholder preference weights with objective distance-based rankings. The suggested technique is verified on modified IEEE 30-, 57-, and 118-bus systems using 30 benchmark instances encompassing deterministic, dynamic, renewable-rich, stressed, and stability-constrained circumstances. A real-world renewable-integrated campus test system at Shivaji University is created to illustrate field-level applicability, as detailed in Cases 31–36. The system depicts the 11 kV distribution network of Shivaji University, Kolhapur, India, incorporating import–export metering, six distribution transformers, and around 930 kW of distributed rooftop solar photovoltaic systems. The campus scenarios analyze grid-only, deterministic photovoltaic systems, stochastic photovoltaic-load interactions, monsoon/cloud effects, evening peak demand, and contingency/reconfiguration operations. In comparison to the grid-only baseline scenario, the optimized photovoltaic-rich campus operation diminishes grid import by as much as 61.33%, energy expenses by 61.10%, grid-associated carbon emissions by 61.31%, active power losses by 29.17%, and voltage magnitude variation by 61.70%. The comparative assessment against MOABC, NSGA-II, MOEA/D, NSGA-III, and MOPSO under uniform parameter configurations demonstrates that the suggested technique attains superior convergence, diversity, feasibility retention, and robustness. In IEEE benchmark systems, it achieves a 3.82% reduction in costs, a 5.47% decrease in emissions, and a 7.31% enhancement in voltage deviation, while DAA increases non-dominated solution retention by around 18%. Hypervolume, inverted generational distance, spacing measure, Friedman ranking, Wilcoxon rank-sum tests, and run-time analysis validate statistical significance and practical computer feasibility. The proposed DS-MOOPF framework provides a scalable and practical optimization platform for renewable-dominant power systems and energy management on higher education campuses.</p> Graphical Abstract <p></p>