Multiobjective optimization framework for renewable energy integration in smart grids with enhanced stability and resilience using reinforcement learning and distributed control systems

Abstract This research develops a multiobjective framework that couples a modified Proximal Policy Optimization agent—augmented with hierarchical experience replay and a parameterized action space—with a trust-aware consensus mechanism in a distributed control layer. Together with a temporal convolutional network–transformer forecaster and an adaptive weighted-sum multiobjective optimizer, this hybrid computational approach was validated on a university microgrid with 45% renewable penetration. Results demonstrated a 37.8% reduction in frequency fluctuations while accommodating 23.6% higher renewable energy penetration, with 42.3% improved resilience during extreme weather events. This framework establishes an effective pathway for accelerating renewable energy adoption.