Akademik Veri Yönetim Sistemi
Journal of Supercomputing, cilt.81, sa.15, 2025 (SCI-Expanded, Scopus)
Linear success history-based adaptive and ensemble sinusoidal parameter adaptation differential evolution (LSHADE-EpSin) algorithm, an improved version of the differential evolution (DE) algorithm, is one of the most powerful algorithms known in the literature and is recommended for problems requiring high-performance computing (HPC). However, in some global optimisation and engineering problems, LSHADE-EpSin can get stuck in local solution traps or fail to achieve a balanced exploration–exploitation performance. To overcome these drawbacks and further improve the performance of the algorithm, in this paper we redesign LSHADE-EpSin with fitness-distance balance (FDB) and natural survivor method (NSM) and propose a new metaheuristic called FDB-NSM-LSHADE-EpSin. By exploiting the properties of both methods, the convergence performance, local optimum avoidance and exploration–exploitation balancing ability of the current algorithm are effectively enhanced. The capability of the proposed FDB-NSM-LSHADE-EpSin is tested on 54 global optimisation problems. According to the statistical analysis results, the average Friedman scores of FDB-NSM-LSHADE-EpSin and LSHADE-EpSin for 54 cases are 1.35 and 1.65, respectively. In the comparison with 24 different competitive algorithms, LSHADE-EpSin ranked 5th and FDB-NSM-LSHADE-EpSin ranked 1st. The proposed algorithm is also tested on 10 different constrained real-world engineering problems. According to the statistical results, the Friedman scores of the proposed algorithm and its competitor are 1.15 and 1.85, respectively. According to the Wilcoxon test pairwise comparison results, FDB-NSM-LSHADE-EpSin outperformed its competitor in 8 out of 10 problems and both algorithms were able to obtain optimal solutions in 2 of them. In addition, the proposed FDB-NSM-LSHADE-EpSin achieved a higher success rate than its competitor in computational complexity analyses.