DEHHO: A Modular HHO Variant with Trend-Guided DE Exploitation and Gaussian-Stochastic Exploration

Abstract This paper presents DEHHO, a modular variant of Harris Hawks Optimization (HHO), tailored for high-dimensional and structurally complex optimization tasks. DEHHO integrates two phase-specific strategies: a stochastic Gaussian perturbation mechanism to boost diversity during exploration and a trend-guided DE/current-to-best/1 update for intensified yet adaptive exploitation. A dynamic balancing scheme further coordinates the use of DE and HHO components to maintain search synergy throughout iterations. Many tests on the CEC2017 and CEC2020 benchmarks demonstrate that DEHHO consistently does better than 18 HHO variants and 8 popular metaheuristics in terms of accuracy, reliability, and efficiency. Ablation studies confirm the individual effectiveness and synergistic contribution of each mechanism, underscoring the framework’s interpretability, modularity, and scalability.