DEA^2H^2: differential evolution architecture based adaptive hyper-heuristic algorithm for continuous optimization

DEA^2H^2: differential evolution architecture based adaptive hyper-heuristic algorithm for continuous optimization

17 May 2024 | Rui Zhong1 · Jun Yu2
This paper introduces a novel hyper-heuristic algorithm, DEA²H², which is based on differential evolution (DE) for solving continuous optimization problems. The algorithm consists of two main components: low-level heuristics (LLHs) and a high-level component. The LLHs are derived from ten DE search operators, while the high-level component incorporates a success-history-based mechanism inspired by the success-history adaptive DE (SHADE) algorithm. If a specific search operator successfully evolves an offspring individual, it is preserved; otherwise, it is replaced by random initialization. The effectiveness of DEA²H² is validated through extensive numerical experiments on CEC2020 and CEC2022 benchmark functions, as well as eight engineering problems. The algorithm is compared against fifteen well-known metaheuristic algorithms, and ablation experiments are conducted to independently evaluate the high-level component. The results demonstrate the superior performance and robustness of DEA²H² across various optimization tasks, highlighting its potential as an effective tool for continuous optimization problems. The source code for this research is available on GitHub.This paper introduces a novel hyper-heuristic algorithm, DEA²H², which is based on differential evolution (DE) for solving continuous optimization problems. The algorithm consists of two main components: low-level heuristics (LLHs) and a high-level component. The LLHs are derived from ten DE search operators, while the high-level component incorporates a success-history-based mechanism inspired by the success-history adaptive DE (SHADE) algorithm. If a specific search operator successfully evolves an offspring individual, it is preserved; otherwise, it is replaced by random initialization. The effectiveness of DEA²H² is validated through extensive numerical experiments on CEC2020 and CEC2022 benchmark functions, as well as eight engineering problems. The algorithm is compared against fifteen well-known metaheuristic algorithms, and ablation experiments are conducted to independently evaluate the high-level component. The results demonstrate the superior performance and robustness of DEA²H² across various optimization tasks, highlighting its potential as an effective tool for continuous optimization problems. The source code for this research is available on GitHub.
Reach us at info@study.space