Metaheuristic algorithms have become effective tools for solving complex engineering optimization problems. The Parrot Optimization (PO) algorithm is a recently proposed method with promising performance; however, it still suffers from premature convergence and slow convergence when handling complex tasks. To address these issues, this paper proposes an improved variant termed the Elite Cauchy Multi-scale Multi-directional Parrot Optimization algorithm (ECMPO). The proposed ECMPO incorporates three coordinated strategies. First, an elite learning strategy is introduced before the behavioral update phase to guide the population toward promising regions and accelerate convergence. Second, after the behavioral update, individuals are adaptively processed based on their fitness: elite individuals perform a multi-scale multi-directional search to enhance local exploitation. Third, non-elite individuals undergo Cauchy mutation to increase population diversity and strengthen global exploration. These strategies enable ECMPO to achieve a better balance between exploration and exploitation. To evaluate its performance, ECMPO is examined through ablation studies on the CEC2017 benchmark functions and further compared with eleven algorithms on the CEC2020–2022 benchmark suites. Finally, it is applied to six constrained engineering design problems. The results demonstrate that ECMPO exhibits superior performance compared with competitive algorithms. ECMPO shows strong robustness and applicability in practical engineering optimization tasks.
Yang et al. (Sun,) studied this question.