Abstract
Evolutionary computation algorithms are widely used for solving complex optimization problems, but their practical performance is often hindered by high computational costs and premature convergence. To address this challenge, we propose an opposition-based chaotic evolution (OBCE) algorithm that integrates opposition-based learning (OBL) into the chaotic evolution framework. The OBL mechanism introduces mirrored solutions to increase population diversity and improve global search ability with minimal overhead. The proposed algorithm is evaluated on a series of single-objective and multi-objective numerical optimization problems, including standard benchmark functions and a real-world hybrid rocket engine design task. Compared to conventional chaotic evolution and other baseline algorithms, OBCE consistently shows faster convergence and better solution quality across various problem types and dimensions. In multi-objective settings, OBCE enhances the diversity of Pareto solutions, offering broader decision-making options. In the rocket engine design task, the algorithm finds more competitive design parameters under realistic constraints. These results demonstrate that integrating OBL into chaotic evolution can effectively mitigate premature convergence and improve optimization performance in both theoretical and applied scenarios. The findings support the broader applicability of OBCE in real-world engineering design problems where solution quality and efficiency are critical.