基于吸引互斥多策略改进雾凇算法

Improved Rime Optimization Algorithm Based on Multi-Strategy of Attraction-Repulsion Mechanism

摘要: 雾凇算法(RIME)是近年提出的一种模拟雾凇生长现象的元启发式优化算法，具有结构简洁、参数较少等优势。但在处理高维复杂的工程优化问题时，该算法仍存在种群多样性缺失、搜索后期易陷入局部最优解以及收敛精度受限等不足。因此，研究如何通过多策略协同改进以提升RIME的全局寻优性能具有重要的理论价值与应用意义。针对上述问题，本文提出一种基于吸引互斥多策略改进的雾凇算法。首先，在算法搜索阶段引入吸引互斥机制，通过模拟粒子间的动态引力与斥力作用来实时调整搜索步长，使个体在搜索前期具备更广的探索范围，在后期实现精准开发，从而有效平衡算法的全局探索与局部搜索能力。其次，借鉴冠豪猪优化算法(CPO)中的防御机制对算法进行二次改进：利用气味防御机制的随机扩散特性增强种群的扰动性，防止算法在迭代后期陷入早熟收敛；同时，通过物理攻击机制的强扰动性能，提升个体在复杂解空间中跳出局部极值的概率。两种策略相互配合，从动力学调整与多样性维护两个维度优化了原算法的生长演化过程。为验证改进算法的性能，选取多组基准测试函数进行仿真实验。结果表明，与原始RIME、CPO及其他典型优化算法相比，改进算法在寻优精度、收敛速度和统计鲁棒性方面均有显著提升。特别是在多峰复杂函数的处理上，该算法展现出卓越的跳出局部最优能力。通过Wilcoxon秩和检验进一步证明了改进策略的统计显著性，验证了所提算法在解决复杂优化任务时的有效性。

Abstract: The Rime Optimization Algorithm (RIME) is a recently proposed metaheuristic algorithm inspired by the physical growth of rime. While it features a simple structure and few parameters, RIME often suffers from insufficient population diversity, premature convergence, and limited accuracy when dealing with complex high-dimensional optimization problems. Therefore, researching how to improve the global optimization performance of RIME through multi-strategy collaborative improvement has important theoretical value and application significance. This paper proposes an improved rime optimization algorithm (ACRIME) based on a multi-strategy coordination approach. First, an attraction-repulsion mechanism is introduced during the algorithm’s search phase. By simulating the dynamic attraction and repulsion between particles, the search step size is adjusted in real time, allowing individuals to have a wider exploration range in the early stages of the search and achieve precise development in the later stages, thus effectively balancing the algorithm’s global exploration and local search capabilities. Second, the algorithm is further improved by borrowing the defense mechanism from the Caucasian porcupine optimization algorithm (CPO): the random diffusion characteristics of the odor defense mechanism are used to enhance the perturbation of the population, preventing the algorithm from falling into premature convergence in the later stages of iteration; at the same time, the strong perturbation performance of the physical attack mechanism is used to increase the probability of individuals escaping local optima in complex solution spaces. The two strategies work together to optimize the growth and evolution process of the original algorithm from two dimensions: dynamic adjustment and diversity maintenance. To verify the performance of the improved algorithm, simulation experiments were conducted using multiple benchmark test functions. The results show that compared with the original RIME, CPO, and other typical optimization algorithms, the improved algorithm has significant improvements in optimization accuracy, convergence speed, and statistical robustness. Especially in the handling of multimodal complex functions, the algorithm exhibits excellent ability to escape local optima. The statistical significance of the improved strategy was further demonstrated by the Wilcoxon rank-sum test, which verifies the effectiveness of the proposed algorithm in solving complex optimization tasks.