Abstract: Chemical Reaction Optimization（CRO） is a recently proposed population-based meta-heuristic, and Real-Coded CRO（RCCRO） is a variant of CRO, as the deficiency of the algorithm, a new Real-Valued CRO（RVCRO）algorithm is formulated and proposed for solving global numerical optimization problems. The new algorithm reformulates the framework and searching paradigm of conventional CRO. It also reduces the number of control parameters from eight to five. Then a set of commonly used benchmark functions is applied to evaluate the performance of the proposed algorithm, and the comparison results indicate that the proposed algorithm can generate superior performance than the RCCRO, in terms of the final error value and convergence speed.

Key words: chemical reaction optimization, real-valued algorithm, meta-heuristic, numerical optimization

摘要： 化学反应优化（CRO）是近年来提出的一个基于种群的元启发式算法，而实数编码化学反应优化（RCCRO）是CRO的一个变体，针对该算法存在的不足提出了一个新的实值化学反应优化（RVCRO）算法求解全局数值优化问题。该算法改进了传统CRO的框架和搜索模式，将控制参数从8个减少到5个;采用一组常用的基准函数来测试其性能，比较结果表明，该算法在最终误差值和收敛速度方面，性能有较大提升。

关键词: 化学反应优化, 实值算法, 元启发式, 数值优化