Designing modern heuristic algorithms to solve the Transmission Expansion Planning problem

Transmission Expansion Planning (TEP) aims at identifying a list of new assets to be installed on the transmission grid to meet the long-term forecasted demand while ensuring a safe supply over the entire planning horizon. As TEP is a Mixed Integer Non-Linear Problem (MINLP) with a huge search space, in the last years several modern heuristic algorithms were proposed to deal with its challenging characteristics. In this way, this paper describes and evaluates the impact and implementation of four operators that can be easily incorporated in any evolutionary algorithm, namely: Neighborhood Search for Local Improvement (NSLI), Diversity Control (DC), Elitist Reproduction (ER) and Boundary Local Search (BLS). The impact of these operators is assessed and discussed over a hundred simulations using a traditional Genetic Algorithm (GA) and a well-known test system, the RTS 24-bus. Regarding the results, the NSLI and the BLS operator considerably improved the GA performance in solving the TEP problem regarding both the final value of the objective function and the diversity of solutions.