School of Physical Education, Northeast Normal University, Changchun 130022, China.
School of Artificial Intelligence, Hebei University of Technology, Tianjin 300401, China.
Math Biosci Eng. 2021 Aug 25;18(6):7143-7160. doi: 10.3934/mbe.2021354.
Chemical Reaction Optimization (CRO) is a simple and efficient evolutionary optimization algorithm by simulating chemical reactions. As far as the current research is concerned, the algorithm has been successfully used for solving a number of real-world optimization tasks. In our paper, a new real encoded chemical reaction optimization algorithm is proposed to boost the efficiency of the optimization operations in standard chemical reactions optimization algorithm. Inspired by the evolutionary operation of the differential evolution algorithm, an improved search operation mechanism is proposed based on the underlying operation. It is modeled to further explore the search space of the algorithm under the best individuals. Afterwards, to control the perturbation frequency of the search strategy, the modification rate is increased to balance between the exploration ability and mining ability of the algorithm. Meanwhile, we also propose a new population initialization method that incorporates several models to produce high-quality initialized populations. To validate the effectiveness of the algorithm, nine unconstrained optimization algorithms are used as benchmark functions. As observed from the experimental results, it is evident that the proposed algorithm is significantly better than the standard chemical reaction algorithm and other evolutionary optimization algorithms. Then, we also apply the proposed model to address the synthesis problem of two antenna array synthesis. The results also reveal that the proposed algorithm is superior to other approaches from different perspectives.
化学反应优化（CRO）是一种通过模拟化学反应来实现的简单而高效的进化优化算法。就目前的研究而言，该算法已成功用于解决许多实际的优化任务。在本文中，我们提出了一种新的实值编码化学反应优化算法，以提高标准化学反应优化算法中的优化效率。受差分进化算法的进化操作的启发，在基本操作的基础上提出了一种改进的搜索操作机制。通过建模进一步探索算法在最优个体下的搜索空间。之后，为了控制搜索策略的扰动频率，我们提高了修正率，以平衡算法的探索能力和挖掘能力。同时，我们还提出了一种新的种群初始化方法，该方法结合了多个模型来生成高质量的初始种群。为了验证算法的有效性，我们使用了 9 个无约束优化算法作为基准函数。从实验结果可以明显看出，所提出的算法明显优于标准化学反应算法和其他进化优化算法。然后，我们还将所提出的模型应用于解决两个天线阵综合问题。结果还表明，从不同角度来看，所提出的算法都优于其他方法。
