Multi-strategy collaborative improvement Dung Beetle Optimization Algorithm for Engineering Problems

Swarm intelligent algorithms have the ability to quickly find optimal solutions to problems, but they suffer from an imbalance between global exploration and local exploitation. The dung beetle optimization (DBO) algorithm was newly developed in 2022 and has excellent comprehensive performance; however, it still suffers from this problem. In this study a new multi-strategy improvement (LCTDBO) is proposed to balance the exploration and exploitation capabilities of the DBO algorithm to avoid falling into the local optimal solution. Latin hypercube sampling initializes the dung beetle to make the population initialization distribution uniform and to search the global space as much as possible. An adaptive nonlinear weight and convergence factor based on the circular arc formula are proposed to improve its global search capability, enhance local exploitation, and accelerate convergence as the number of iterations increases. Applying the t-distribution mutation strategy mutates the optimal solution of each iteration to increase the population diversity and avoid local optimal solutions. This research utilizes 23 classic benchmark functions and CEC-2022 to compare LCTDBO with six classic swarm intelligent algorithms, and three extensive practical engineering problems to verify LCTDBO's ability to solve practical problems. The results show that LCTDBO balances the global search and local exploitation well, accelerates algorithm convergence, and has excellent performance in solving practical engineering problems.

群智能算法具备迅速寻找到问题最优解的能力，然而，其在全局探索与局部利用之间存在着不均衡。2022年新近开发的粪便甲虫优化（DBO）算法展现出卓越的综合性能；然而，该算法亦面临着此等问题。本研究提出了一种新型的多策略改进方法（LCTDBO），旨在平衡DBO算法的探索与利用能力，以规避陷入局部最优解的困境。拉丁超立方抽样技术被用于初始化粪便甲虫，旨在实现种群初始化分布的均匀性，并尽可能地在全局空间中进行搜索。基于圆弧公式的自适应非线性权重及收敛因子被提出，以提升算法的全局搜索能力，增强局部利用效率，并随着迭代次数的增加加速收敛。应用t分布变异策略对每轮迭代的最优解进行变异，以增加种群多样性并避免局部最优解。本研究采用23个经典基准函数以及CEC-2022进行测试，将LCTDBO与六种经典群智能算法进行对比，并针对三个广泛应用的工程实际问题进行验证，以测试LCTDBO解决实际问题的能力。研究结果揭示了LCTDBO在全局搜索与局部利用之间的平衡能力，其加速了算法的收敛速度，并在解决实际工程问题方面展现出卓越的性能。