文章聚焦于蜣螂优化算法所固有的局限性问题,具体表现为其易于陷入局部最优解、在全局搜索能力上有所欠缺,以及收敛速度相对缓慢。针对这些不足,文章提出了一种创新性的改进策略——多策略融合的改进型蜣螂优化算法(简称MSIDBO)。在该改进方案中,首先于算法的初始化阶段引入了Logistic混沌映射机制,旨在有效提升种群分布的均匀程度;其次,采用鱼鹰优化算法替换原有蜣螂算法中的滚球位置更新机制,以解决原算法仅依赖最差值进行位置更新、缺乏个体间即时交流及参数冗余的问题;最后,实施了自适应t分布扰动策略,旨在迭代初期强化全局探索能力,而在迭代末期则加强局部搜索效率,并加速了算法的收敛进程。为了验证MSIDBO算法的有效性,对14个经典测试函数和工程应用问题进行测试,结果表明,引入的3种策略能有效提升蜣螂优化算法的性能。This study examines the intrinsic limitations of the dung beetle optimization algorithm, particularly its propensity to converge on local optima, its insufficient global search capabilities, and its relatively slow convergence rate. To mitigate these issues, the paper introduces a novel enhancement strategy termed the Improved Dung Beetle Optimization Algorithm with Multi-Strategy Fusion (MSIDBO). This enhancement involves several key modifications: first, a Logistic Chaos mapping mechanism is incorporated during the initialization phase of the algorithm to enhance the uniformity of population distribution. Second, the Fishhawk optimization algorithm is employed to replace the original rolling ball position update mechanism of the dung beetle algorithm. This substitution addresses the original algorithm’s reliance on the worst value for position updates, the absence of instantaneous communication among individuals, and the presence of parameter redundancy. Lastly, an adaptive t-distribution perturbation strategy is introduced to bolster global exploration during the initial iterations while simultaneously improving local search efficiency in the later stages, thereby accelerating the overall convergence of the algorithm. To evaluate the efficacy of the MSIDBO algorithm, a series of tests involving 14 classical benchmark functions and engineering application problems were conducted. The results indicate that the three strategies implemented significantly enhance the performance of the dung beetle optimization algorithm.
蜣螂优化算法(Dung Beetle Optimizer, DBO)是Xue等人在2022年提出的一种新的群体智能优化算法,其灵感来源于蜣螂的生物行为过程。针对蜣螂优化算法全局探索和局部开发能力不平衡、容易陷入局部最优等缺点,提出了一种混合策略改进的蜣螂优化算法(MIDBO)。首先,在种群初始化时,引入Tent混沌反向学习策略,使初始种群成员能够均匀分布,增加种群丰富性;其次,引入三角形随机游走策略改进繁殖蜣螂位置更新方式,平衡了全局搜索和局部挖掘能力;然后,加入动态权重系数改进蜣螂偷窃行为,加快算法的收敛速度;最后,引入混合变异算子对最优蜣螂位置进行扰动,提高算法跳出局部最优的能力。将所提算法与其他知名优化算法进行了15个基准测试函数的测试比较,仿真结果表明,MIDBO算法是可行有效的,其寻优精度和收敛速度都有了很大的提高,总体性能更好。Dung Beetle Optimizer (DBO) is a new swarm intelligence optimization algorithm proposed by Xue et al. in 2022, inspired by the biological behavior process of dung beetles. A mix-strategy improved dung beetle optimizer (MIDBO) is proposed to address the drawbacks of imbalanced global exploration and local development capabilities, as well as the tendency to fall into local optima. Firstly, during population initialization, a Tent chaotic reverse learning strategy is introduced to enable the initial population members to be evenly distributed and increase population richness;secondly, the introduction of triangle random walk strategy improves the position update method of breeding dung beetles, balancing global search and local mining capabilities;then, a hybrid mutation operator is adopted to improve the theft behavior of dung beetles and accelerate the convergence speed of the algorithm;finally, a mixed mutation operator is introduced to perturb the optimal dung beetle position, improving the algorithm’s ability to jump out of local optima. The proposed algorithm was compared with other well-known optimization algorithms through 15 benchmark test functions, and simulation results showed that the MIDBO algorithm is feasible and effective. Its optimization accuracy and convergence speed have been greatly improved, and the overall performance is better.
蜣螂优化算法(DBO)作为一种新兴的智能优化算法,在求解复杂优化问题中显示出巨大潜力。然而,其在收敛精度和易陷入局部最优方面的局限性限制了其应用范围。本文提出了一种多策略改进的蜣螂优化算法(ODBO),通过佳点集群初始化和周期突变机制增强种群多样性,引入Beta分布动态生成反射解决方案以探索更广的搜索空间,并采用莱维飞行处理边界违规问题。进一步融合鲸鱼优化算法的螺旋搜索更新机制,结合随机策略更新位置,显著提升了算法的收敛精度和鲁棒性。当算法陷入停滞时,引入t分布扰动变异策略,有效提高了算法跳出局部最优解的能力。通过在17个基准测试函数验证了改进策略的有效性。此外,本文还将ODBO应用于车间排产调度问题,进一步证实了其在解决实际工程问题中的有效性和可靠性。The Dung Beetle Optimization algorithm (DBO), as an emerging intelligent optimization algorithm, shows great potential in solving complex optimization problems. However, its limitations in convergence precision and susceptibility to local optima hinder its broader application. This paper proposes an improved multi-strategy Dung Beetle Optimization algorithm (ODBO), which enhances population diversity through good point set initialization and periodic mutation mechanisms. A Beta distribution is introduced to dynamically generate reflected solutions to explore a wider search space, and Lévy flight is applied to handle boundary violation issues. Additionally, the spiral search update mechanism from the Whale Optimization Algorithm is integrated, along with random strategy updates for position, significantly improving the algorithm’s convergence accuracy and robustness. When the algorithm stagnates, a t-distribution mutation perturbation strategy is introduced, effectively enhancing its ability to escape local optima. Simulations on 17 benchmark functions test functions validate the effectiveness of the improved strategies. Moreover, the application of ODBO to the job-shop scheduling problem confirms its effectiveness and reliability in solving real-world engineering problems.
暂无评论