Ma Yunpeng, Meng Wanting, Wang Xiaolu, Gu Peng, Zhang Xinxin
School of Information Engineering, Tianjin University of Commerce, Beichen, Tianjin 300134, China.
College of Science, Tianjin University of Commerce, Beichen, Tianjin 300134, China.
Biomimetics (Basel). 2025 May 8;10(5):299. doi: 10.3390/biomimetics10050299.
The Sparrow Search Algorithm (SSA), proposed by Jiankai Xue in 2020, is a swarm intelligence optimization algorithm that has received extensive attention due to its powerful optimization-seeking ability and rapid convergence. However, similar to other swarm intelligence algorithms, the SSA has the problem of being prone to falling into local optimal solutions during the optimization process, which limits its application effectiveness. To overcome this limitation, this paper proposes a Modified Sparrow Search Algorithm (MSSA), which enhances the algorithm's performance by integrating three optimization strategies. Specifically, the Latin Hypercube Sampling (LHS) method is employed to achieve a uniform distribution of the initial population, laying a solid foundation for global search. An adaptive weighting mechanism is introduced in the producer update phase to dynamically adjust the search step size, effectively reducing the risk of the algorithm falling into local optima in later iterations. Meanwhile, the cat mapping perturbation and Cauchy mutation operations are integrated to further enhance the algorithm's global exploration ability and local development efficiency, accelerating the convergence process and improving the quality of the solutions. This study systematically validates the performance of the MSSA through multi-dimensional experiments. The MSSA demonstrates excellent optimization performance on 23 benchmark test functions and the CEC2019 standard test function set. Its application to three practical engineering problems, namely the design of welded beams, reducers, and cantilever beams, successfully verifies the effectiveness of the algorithm in real-world scenarios. By comparing it with deterministic algorithms such as DIRET and BIRMIN, and based on the five-dimensional test functions generated by the GKLS generator, the global optimization ability of the MSSA is thoroughly evaluated. In addition, the successful application of the MSSA to the problem of robot path planning further highlights its application advantages in complex practical scenarios. Experimental results show that, compared with the original SSA, the MSSA has achieved significant improvements in terms of convergence speed, optimization accuracy, and robustness, providing new ideas and methods for the research and practical application of swarm intelligence optimization algorithms.
麻雀搜索算法（SSA）由薛建凯于2020年提出，是一种群体智能优化算法，因其强大的寻优能力和快速收敛性而受到广泛关注。然而，与其他群体智能算法类似，SSA在优化过程中存在容易陷入局部最优解的问题，这限制了其应用效果。为克服这一局限性，本文提出了一种改进的麻雀搜索算法（MSSA），通过整合三种优化策略来提升算法性能。具体而言，采用拉丁超立方抽样（LHS）方法实现初始种群的均匀分布，为全局搜索奠定坚实基础。在生产者更新阶段引入自适应加权机制，动态调整搜索步长，有效降低算法在后续迭代中陷入局部最优的风险。同时，集成猫映射扰动和柯西变异操作，进一步增强算法的全局探索能力和局部开发效率，加速收敛过程并提高解的质量。本研究通过多维实验系统地验证了MSSA的性能。MSSA在23个基准测试函数和CEC2019标准测试函数集上展现出优异的优化性能。将其应用于焊接梁、减速器和悬臂梁这三个实际工程问题，成功验证了该算法在实际场景中的有效性。通过与DIRET和BIRMIN等确定性算法进行比较，并基于GKLS生成器生成的五维测试函数，全面评估了MSSA的全局优化能力。此外，MSSA在机器人路径规划问题上的成功应用进一步凸显了其在复杂实际场景中的应用优势。实验结果表明，与原始SSA相比，MSSA在收敛速度、优化精度和鲁棒性方面均取得了显著提升，为群体智能优化算法 的研究和实际应用提供了新的思路和方法。
