Computer Engineering and Applications ›› 2022, Vol. 58 ›› Issue (9): 279-287.DOI: 10.3778/j.issn.1002-8331.2011-0378

• Engineering and Applications • Previous Articles     Next Articles

Mix with Multiple Strategies Bird Swarm Algorithm and Optimization of ELM Model in Oil Layer Classification

SONG Fei, XIA Kewen, YANG Wenbiao   

  1. School of Electronic and Information Engineering, Hebei University of Technology, Tianjin 300401, China
  • Online:2022-05-01 Published:2022-05-01



  1. 河北工业大学 电子信息工程学院,天津 300401

Abstract: In order to improve the shortcomings of bird swarm algorithm, such as easy to fall into local optimum, slow convergence speed and insufficient diversity of population, the mix with multiple strategies bird swarm algorithm is proposed. The chaos weight, symmetric tangent chaos acceleration coefficient and Gaussian disturbance strategy are introduced to enhance the ability of the algorithm to jump out of the local optimum. The hybrid multi-step selection and adaptive step size factor strategy are introduced to accelerate the convergence speed of the algorithm. The wavelet mutation strategy is introduced to increase the population diversity of the algorithm. 10?benchmark functions are used in the experiment, and the improved algorithm is compared with the other five intelligent algorithms, and the performance of the improved algorithm is verified to be better than other algorithms. In addition, in order to improve the accuracy of extreme learning machine(ELM) model in oil layer classification, the improved bird swarm algorithm is used to optimize the parameters of ELM model. The actual logging application shows that the ELM model optimized by improved bird colony algorithm has a significant effect in reservoir identification, which is better than ELM model optimized by genetic algorithm, particle swarm optimization algorithm and ant colony algorithm.

Key words: bird swarm algorithm, extreme learning machine, oil layer classification

摘要: 为了改进鸟群算法易陷入局部最优、收敛速度慢以及种群多样性不足的缺点,提出融合多策略的鸟群算法。引入混沌权重和对称切线混沌加速系数以及高斯扰动策略,增强算法跳出局部最优的能力;引入混合多步选择和自适应步长因子策略,加快算法的收敛速度;引入小波变异策略,丰富算法的种群多样性。实验采用10个基准测试函数,将改进的算法与另外5种智能算法进行仿真对比,验证了改进的算法性能优于其他算法。另外,为了提高极限学习机(ELM)在油层识别中的精度,将改进的鸟群算法用于ELM模型的参数优化。实际测井应用表明基于改进鸟群算法优化的ELM模型在油层识别中效果显著,优于基于遗传算法、粒子群算法、蚁群算法优化的ELM模型。

关键词: 鸟群算法, 极限学习机, 油层识别