改进细菌觅食算法在高维优化问题中的应用

针对以往细菌觅食优化算法自适应步长公式经验性参数过多、无法真正实现自适应的缺点,提出了改进的步长公式,使步长仅与细菌个体当前的进化代数和所求解问题的寻优范围有关,真正实现步长的自适应；其次,将混沌思想和差分进化思想与细菌觅食算法结合,对算法初始化过程和寻优过程进行改进,增加群体多样性,避免算法因为早熟而陷入局部最优值；在高维问题的优化过程中,采用逐维更新细菌位置的方法,将整体问题分维处理,极大地提高了算法效率和精度。通过对多个标准测试函数在多维空间进行测试,表明改进算法在高维空间中寻优时速度快、精度高、求解过程简单可行,在寻得最优解的精度上比其他改进方案有显著提高。

Application of Bacteria Foraging Algorithm in High-dimensional Optimization Problems

Excessive empirical parameters in the former self-adaptation step formula caused the defect in terms of failing to achieve self-adaptation in bacteria foraging optimization algorithm.Therefore,a revised step formula has been proposed,which enables step length to be relevant to the present evolution generation of individual bacteria as well as the optimal range of the problem to be solved,in order to achieve the step length self-adaption.Besides,the combination of chaotic thoughts and differential evolution thoughts with bacteria foraging algorithm can improve both the initial process and optimal process of the algorithm.This method increased the diversity of groups,preventing the algorithm from falling into the local optima due to the precocious.In the optimal process of high-dimensional problem,fractal dimension optimization is used to replace the former method.The fractal dimension optimization means that the information of every dimension will be updated one by one on the basis of whether the new position of every dimension changes comparing to the fitness value.Dealing with the problems in different dimensions can boost the precision and efficiency of the algorithm obviously.Experiments show that through the testing of multiple standard test functions in the hyperspace,the revised algorithm optimizing in the hyperspace has several benefits,such as fast speed,high precision and the simple process of solving.It has improved manifestly in terms of precision when comparing to other modified programs.