一种基于灰预测理论的混合蛙跳算法

Shuffled Frog Leaping Algorithm Based on Grey Prediction Theory

为提高混合蛙跳算法在优化问题求解中的性能,提出一种基于灰预测理论的改进混合蛙跳算法。该算法首先将基本算法的进化模式进行调整,强化了进化过程中全局信息的交换;之后引入移动步长变异算子,根据进化过程的不同阶段和利用灰预测理论获得进化过程中最优解进步速度,并借鉴模糊控制思想对该变异算子进行控制,进而实现移动步长的自适应调整。采用6个标准测试函数,与基本算法和已有改进算法进行性能对比分析,证明了改进后的混合蛙跳算法在收敛精度、收敛速度和收敛成功率方面的优越性及灰预测理论在算法改进领域中的可行性。最后,将改进算法应用于10 k V油浸式配电变压器优化设计工作中,验证了该改进算法的实用性。

To enhance the performance of shuffled frog leaping algorithm in solving optimization problems,a new model for hybrid leapfrog algorithm based on grey prediction theory was proposed. The algorithmic evolution model was adjusted to strengthen the ability to exchange the global information in the process of evolution. Then the algorithm implemented the mobile step self-adaption adjustment through introduced mobile step mutation operator. The mutation operator was controlled by the different stages of evolution and the optimal solution progress speed in the process of evolution obtained by grey prediction theory and the fuzzy control thoughts. The advantages of the improved hybrid leapfrog algorithm,such as the accuracy,convergent speed and success rate,and the feasibility of grey prediction theory in the field of algorithm improvement,is verified by comparison with the basic shuffled frog leaping algorithm and the known improved algorithm on performance through six standard test functions. Finally,the practicability of the improved algorithm is proved by applying it to 10 k V oil-immersed distribution transformer optimization design works.