Abstract
This paper introduces a variant of differential evolution with micro-populations, called μ-DE-ERM, which incorporates a periodic elitist replacement mechanism with the aim of preserving diversity without the need to measure it explicitly. The proposed algorithm is designed for scenarios with reduced evaluation budgets, where efficiency and convergence stability are critical. Its performance is evaluated on CEC 2005 and CEC 2017 benchmark suites, covering unimodal, multimodal, hybrid, and composition functions, as well as on two real-world engineering problems: the identification of dynamic parameters and the tuning of a PID controller for a one-degree-of-freedom robotic manipulator. The comparative analysis shows that μ-DE-ERM achieves competitive or superior results against its predecessors DE and μ-DE, and remains effective when contrasted with advanced algorithms such as L-SHADE and RuGA. Furthermore, additional comparisons with algorithms with competitive replacement mechanisms, μ-DE-Cauchy and μ-DE-Shrink, confirm the robustness of the proposal in real applications, particularly under strict computational constraints. These findings support μ-DE-ERM as a practical and efficient alternative for optimization problems in resource-limited environments, delivering reliable solutions at low computational cost.