A reactive GRASP for transmission network expansion planning

In this paper we describe an enhanced GRASP (Greedy Randomized Adaptive Search Procedure) applied to power transmission network expansion planning problems. GRASP is a metaheuristic that has been shown to be powerful in solving combinatorial problems. It is composed of two phases: the construction phase where a feasible solution is iteratively built, and a local search phase that seeks improvements within a given neighborhood of the solution found by the construction phase. The best solution over all GRASP iterations is chosen as the result. The enhancements analyzed in this paper are two; instead of using a random function in the construction phase, we use a linear bias function to guide the construction. We also applied a reactive procedure to self adjust GRASP parameters. Two real-world power transmission network expansion planning problems of the Brazilian power system are used to verify the performance of these improvements.