一种求解多维全局优化问题的改进蝙蝠算法

Improved Bat Algorithm for Solving Multi-dimensional Global Optimization Problems

蝙蝠算法是一种模拟蝙蝠回声定位行为的新型群智能优化算法,对多维函数,个体在全局最佳蝙蝠的引导下修改所有的维,这种候选解生成方式可能导致种群多样性下降过快和算法局部求精能力不足.针对这些不足,提出一种改进的蝙蝠算法,使用随机蝙蝠来引导个体飞行和局部搜索,以提高种群多样性,使用修改部分维的策略来加强算法的局部求精能力.在典型测试函数上对新算法进行了仿真,结果表明改进的蝙蝠算法能够有效提高算法的收敛速度并改善解的质量,与其它改进蝙蝠算法和改进群智能算法的比较表明,改进算法在求解多维函数优化问题上是具有竞争力的.

Bat algorithm （ BA ） is a new swarm intelligence optimization algorithm, which simulates the echolocation behavior of bats. For multi-dimensional function, individual produces candidate solution by modifying all dimensions under the guidance of the current best bat, which may decline the swarm diversity rapidly and deteriorate its local searching ability. To tackle those shortages, this paper proposes an improved bat algorithm. To enhance the swarm diversity, randomly selected bats are used to guide the flight and the local searching. To improve the local searching ability, bats modify part dimensions only when they fly and process local searching. The sim- ulation experiments on typical test functions show the proposed method can improve the convergence speed and the quality of the solu- tion effectively. Meanwhile, the results also reveal the proposed algorithm is competitive for solving multi-dimensional function optimi- zation compared with other improved bat algorithm and swarm intelligence algorithms.