基于分子动力学模拟的改进混合蛙跳算法

Improved Shuffled Frog Leaping Algorithm Based on Molecular Dynamics Simulations

针对基本的混合蛙跳算法(Shuffled frog leaping algorithm,SFLA)后期搜索速度变慢,容易陷入局部最优解的缺点,借鉴分子动力学(Molecular dynamics,MD)模拟的思想,提出一种基于分子动力学模拟的改进的混合蛙跳算法。该算法将种群中的粒子等效成分子,并提出一种新的分子间作用力计算方法来代替两体间经典的Lennard-Jones作用力计算方法,利用Velocity-Verlet算法和高斯变异算子代替基本混合蛙跳算法的更新策略,有效地平衡了种群的多样性和搜索的高效性。高维多峰函数测试的结果表明,基于分子动力学模拟的改进混合蛙跳算法能提高算法后期跳出局部极值的能力,全局寻优能力明显优于基本的混合蛙跳算法。

To overcome the defects of shuffled frog leaping algorithm （SFLA） such as slow searching speed in the late evolution and easily trapping into local extremum, an improved shuffled frog leaping algorithm （ISFLA） based on the basic ideas of molecular dynamics （MD） simulations is presented with the population being regarded as a molecular system. A new intermolecular force is proposed instead of the classic two-body Lennard-Jones force and Velocity-Verier algorithm and Gaussian mutation are adopted to replace the original SFLA update strategy. So the population diversity and search efficiency can be effectively balanced. Test results on high-dimensional and multi-modal optimization problems indicate that ISFLA improves the capacity of escaping from local maximum and the global searching performance is superior to SFLA.