LAUSR

Nature-inspired optimization algorithms have received more and more attention from the researchers due to their several advantages. Genetic algorithm (GA) is one of such bio-inspired optimization techniques, which has mainly three operators, namely selection, crossover, and mutation. Several attempts had been made to make these operators of a GA more efficient in terms of performance and convergence rates. In this paper, a directional crossover (DX) has been proposed for a real-coded genetic algorithm (RGA). As the name suggests, this proposed DX uses the directional information of the search process for creating the children solutions. Moreover, one method has been suggested to obtain this directional information for identifying the most promising areas of the variable space. To measure the performance of an RGA with the proposed crossover operator (DX), experiments are carried out on a set of six popular optimization functions, and the obtained results have been compared to that yielded by the RGAs with other well-known crossover operators. RGA with the proposed DX operator (RGA-DX) has outperformed the other ones, and the same has been confirmed through the statistical analyses. In addition, the performance of the RGA-DX has been compared to that of other five recently proposed optimization techniques on the six test functions and one constrained optimization problem. In these performance comparisons also, the RGA-DX has outperformed the other ones. Therefore, in all the experiments, RGA-DX has been found to yield the better quality of solutions with the faster convergence rate.