Teaching quality evaluation-based differential evolution and its application on synthesis of linear sparse arrays

To enhance the solution quality and accelerate the convergence speed, teaching quality evaluation-based differential evolution algorithm with Levy flight and adaptive parameters is proposed in this paper, which contains two modifications. One is the new multi-population strategy inspired by teaching quality evaluation. In this multi-population strategy, first, the entire evolving population is partitioned into several subpopulations of equal size and each subpopulation is evaluated and assigned an evaluation score like the way in teaching quality evaluation, according to which new individuals are generated. Then the subpopulations integrate and are re-divided to make some individuals in the better subpopulations spread to other subpopulations, thus guiding the evolution of the entire population and accelerating the convergence speed. The other modification is the improved mutation strategy combined with Levy flight and adaptive parameters. Levy flight can avoid the algorithm stuck with the local optimum, and adaptive parameters can change dynamically during the evolutionary process, hence reducing the adverse effects caused by parameter sensitivity. To validate the effectiveness of the proposed algorithm, extensive test experiments on benchmark functions are carried out, and the two modifications are studied separately at length. At the end, the proposed algorithm is applied to synthesize linear sparse array antennas. The results show that it can reduce peak sidelobe level further and enhance the performance of linear sparse arrays.