基于t-分布的混沌蝙蝠算法及其在弹簧设计中的应用

Chaotic Bat Algorithm Based on T-distribution and Its Application in Spring Design

蝙蝠算法(BA)作为一种新型的元启发算法,应用于解决复杂系统的优化问题。但由于算法后期收敛速度较慢,容易产生陷入局部最优的问题。采用Tent混沌映射的随机性和遍历性对种群进行初始化,可有效提升种群分布均匀度,进而通过动态递减的惯性权重更新速度公式,并在局部搜索过程中对非优蝙蝠个体进行自适应t-分布变异,在此基础上提出改进的蝙蝠算法(CTBA),有效规避了算法在搜索过程中的早熟收敛,提高了算法的探索和开发能力。模拟实验选择8个基准测试函数,与其他算法进行对比并且应用于弹簧设计问题,结果表明CTBA算法的性能更优。

Bat algorithm(BA)is a new meta heuristic algorithm,which is applied to solve the optimization problems of complex systems.However,due to the slow convergence speed in the later stage of the algorithm,it is easy to fall into local optimization.The randomness and the ergodicity of tent chaotic map are used to initialize the population and improve the uniformity of population distribution.Then,the speed formula is updated by dynamically decreasing inertia weight,and the non-optimal bat individuals are mutated by adaptive t-distribution in the local search process.On this basis,a corrected bat algorithm(CTBA)is proposed,which effectively avoids the premature convergence of the algorithm in the search process and improves the exploration and development ability of the algorithm.Compared with BA and ABA algorithms,eight test functions are selected for comparative experiments and the application of spring design problems.The results show that the performance of the proposed CTBA is better.