An evolutionary multiobjective based approach to improve robustness in directional overcurrent relay coordination

The coordination of electrical power system protection relays is a hard problem to solve, due to its discrete/nonlinear nature and its complex constraint structure. A multiobjective algorithm to coordinate directional overcurrent relays is proposed in this work. Two objective functions are considered: minimization of the sum of relay operation times, and; maximization of the minimum coordination time interval between relay pairs (primary and backup). The second objective is novel, and we prove to be very useful to improve the coordination robustness. The search for efficient solutions is performed by a matheuristic algorithm, which combines Non-dominated Sorting Genetic Algorithm II, Differential Evolution and Mathematical Programming formulations. Results for four case studies from the literature suggest that the proposed approach is able to generate solutions that are equivalent, or even better, than the ones obtained with mono-objective approaches. In addition, the proposed algorithm can find some solutions that are considerably more robust taking into account variations in system conditions.