CAJ | 학술논문

从混合蛙跳算法的寻优原理出发，研究了其寻优机制。针对标准算法中存在的初始种群不均匀、迭代后期收敛速度慢，易陷入局部最优的缺陷，提出了一种改进的混合蛙跳算法。采用随机化均匀设计方法产生初始种群；引入影响因子，动态地改变子群当前最差值对其进化行为的影响；根据群体适应度方差判断种群是否陷入局部最优，并通过对当前全局最优值微扰，使算法跳出局部最优。实验结果表明，改进算法有更高的收敛精度和更好的收敛结果。
종혼합와도산법적심우원리출발，연구료기심우궤제。침대표준산법중존재적초시충군불균균、질대후기수렴속도만，역함입국부최우적결함，제출료일충개진적혼합와도산법。채용수궤화균균설계방법산생초시충군；인입영향인자，동태지개변자군당전최차치대기진화행위적영향；근거군체괄응도방차판단충군시부함입국부최우，병통과대당전전국최우치미우，사산법도출국부최우。실험결과표명，개진산법유경고적수렴정도화경호적수렴결과。
Based on the principle of the shuffled frog leaping algorithm (SFLA), the algorithm’s optimization mechanism is studied. A novel shuffled frog leaping algorithm is proposed to solve the problems of the original SFLA,such as non-uniform initial population,slow convergence speed in later iterations,and being easy to fall into local optimum. The improved algorithm generates the initial population with the random uniform design method,changes the influence of the subgroup’s current worst value on the subsgroup’s evolution behavior dynamically by using the influence factor.Besides,the variance of the population's fitness is calculated to j udge whether the population falls into local optimum,and then the improved algorithm makes the population j ump out of local optimal state by the perturbation of the current global optimal value.Experimental results show that the improved algorithm can lead to a higher convergence accuracy and a better convergence result.