An enhanced adaptive elephant herding optimization based on hybrid cuckoo search algorithm and elite opposition-based learning

This study proposes a hybrid variant of Elephant Herding Optimization that combines an adaptive clan-updating schedule, Lévy-flight exploration from Cuckoo Search, and elite opposition-based sampling. The design targets three known limitations-premature convergence, imbalance between exploration and exploitation, and slow late-stage progress-by (i) nonlinearly annealing clan influence across iterations, (ii) applying long-range moves exclusively to clan leaders, and (iii) injecting diversity around elite candidates. On ten standard benchmarks, the proposed method achieves faster convergence and lower error than EHO, PSO, SCA, and EHO-based hybrids.For example, it reaches the global optimum on F6 and attains a mean error on F1 that is 12.8× lower than that of PSO under identical evaluation budgets. A signal-processing case study (Wiener spline filter design) further demonstrates transferability to an industry-relevant task. We analyze parameter sensitivity, provide a time-complexity breakdown, and discuss limitations. The approach offers a systematic and generalizable hybrid framework that balances broad exploration early with precise exploitation late.