A Two-Stage Hybrid GA-PSO Algorithm for Adaptive Overcurrent Protection in Dynamic Distribution Systems

Abstract

Adaptive overcurrent protection in active distribution networks becomes challenging when distributed energy resources and topology reconfigurations alter fault-current magnitude, direction, and relay coordination relationships. Conventional single-stage heuristic methods may suffer from premature convergence or constraint paralysis when coordination time interval, sensitivity, and thermal constraints must be satisfied simultaneously. This paper proposes a Two-Stage Hybrid Genetic Algorithm and Particle Swarm Optimization framework for adaptive directional overcurrent relay coordination. The first stage uses GA with moderate penalties for global exploration, while the second stage uses PSO with stricter penalties for local refinement of coordination margins and instantaneous high-set current. By treating the high-set current as an explicit decision variable together with pickup current and time multiplier setting, the proposed method directly addresses fatal squeeze, where primary and backup relays may operate instantaneously at the same time. Validation on a modified radial IEEE 14-bus distribution system shows improved convergence behavior while maintaining relay sensitivity, thermal security, and selectivity in the evaluated DER and topology variation scenarios.