Computer Engineering and Applications ›› 2021, Vol. 57 ›› Issue (22): 166-176.DOI: 10.3778/j.issn.1002-8331.2103-0491

• Pattern Recognition and Artificial Intelligence • Previous Articles     Next Articles

Improved Bat Algorithm Based on Self-adaptive Doppler and Dynamic Neighborhood Strategy

CHEN Yao, CHEN Si   

  1. 1.School of Science, Xijing University, Xi’an 710123, China
    2.School of Natural and Applied Sciences, Northwestern Polytechnical University, Xi’an 710072, China
  • Online:2021-11-15 Published:2021-11-16



  1. 1.西京学院 理学院,西安 710123
    2.西北工业大学 理学院,西安 710072


Bat Algorithm(BA) is a new type of meta-heuristic algorithm. Aiming at the problems of reduced optimization accuracy and easy to trap into local optimum, an improved algorithm with adaptive Doppler strategy and dynamic neighborhood strategy SDDNBA is proposed. According to the relative distance between the bats and the prey during the predation process, the adaptive Doppler strategy is introduced to improve the parameter of frequency and enhance the optimization ability of the algorithm for global exploration. Meanwhile, the dynamic neighborhood strategy is combined with the BA. This strategy can increase the diversity of the algorithm’s optimization structure and avoid the algorithm falling into the local optimum. The convergence and computational complexity of the SDDNBA algorithm are theoretically analyzed. In the numerical experiment part, comparative experiments are carried out on the improved SDDNBA algorithm, and numerical simulation comparative experiments are carried out on 10 classic test functions in different dimensions. In addition, it is applied to the optimization design problem of coil compression spring and compared with other algorithms. The results fully prove the effectiveness of the improved algorithm, with better convergence speed, convergence accuracy and stability robustness.


Key words: swarm intelligence optimization algorithm, Bat Algorithm(BA), adaptive strategy, dynamic neighborhood


蝙蝠算法(Bat Algorithm,BA)是一类新型元启发式算法,针对其在算法后期寻优精度降低、易陷入局部极值的不足,提出一种具有自适应多普勒策略及动态邻域策略的改进算法。根据蝙蝠个体在捕食过程中与猎物间存在的相对运动现象,引入自适应多普勒策略改进频率参数,增强算法全局探索的寻优能力。将动态邻域策略与BA算法有机结合,增加蝙蝠个体寻优结构的多样性,改善算法易陷入局部最优的不足。从理论上分析了改进后算法的收敛性和运算复杂性。在数值实验部分对改进后的算法进行了性能及应用测试:对10个经典标准测试函数在不同维度下进行对比实验,将其应用于求解螺旋压缩弹簧优化设计问题,并与其他算法进行了对比分析。实验结果证明了具有自适应多普勒策略及动态邻域策略的改进算法具有更优的收敛速度、收敛精度以及稳定鲁棒性。

关键词: 群智能优化算法, 蝙蝠算法(BA), 自适应策略, 动态邻域