计算机技术与发展
計算機技術與髮展
계산궤기술여발전
Computer Technology and Development
2015年
11期
67-70
,共4页
混合蛙跳算法%种群分割%学习策略%K最短路径
混閤蛙跳算法%種群分割%學習策略%K最短路徑
혼합와도산법%충군분할%학습책략%K최단로경
Shuffled Frog Leaping Algorithm (SFLA)%segmentation population%learning strategy%K shortest path
种群分割是混合蛙跳算法的重要组成部分,不同的种群分割方法对混合蛙跳算法的收敛速度的影响不同.文中首先在原始混合蛙跳算法基础上,提出一种新的种群分割方法,使得每个族群中的个体适应度趋近均衡.然后结合Yen算法的偏离路径思想提出一种新的学习策略,对算法迭代方法进行改进.改进后的混合蛙跳算法适用于求解K最短路径问题.最后对改进后的算法进行仿真实验.首先选取单调递归的Dijkstra算法对改进算法可行性进行验证,结果表明改进后的算法是可行的;再选取遗传算法与改进算法进行对比,结果表明改进后的算法在寻优精确度、时间效率和稳定性方面均优于遗传算法.
種群分割是混閤蛙跳算法的重要組成部分,不同的種群分割方法對混閤蛙跳算法的收斂速度的影響不同.文中首先在原始混閤蛙跳算法基礎上,提齣一種新的種群分割方法,使得每箇族群中的箇體適應度趨近均衡.然後結閤Yen算法的偏離路徑思想提齣一種新的學習策略,對算法迭代方法進行改進.改進後的混閤蛙跳算法適用于求解K最短路徑問題.最後對改進後的算法進行倣真實驗.首先選取單調遞歸的Dijkstra算法對改進算法可行性進行驗證,結果錶明改進後的算法是可行的;再選取遺傳算法與改進算法進行對比,結果錶明改進後的算法在尋優精確度、時間效率和穩定性方麵均優于遺傳算法.
충군분할시혼합와도산법적중요조성부분,불동적충군분할방법대혼합와도산법적수렴속도적영향불동.문중수선재원시혼합와도산법기출상,제출일충신적충군분할방법,사득매개족군중적개체괄응도추근균형.연후결합Yen산법적편리로경사상제출일충신적학습책략,대산법질대방법진행개진.개진후적혼합와도산법괄용우구해K최단로경문제.최후대개진후적산법진행방진실험.수선선취단조체귀적Dijkstra산법대개진산법가행성진행험증,결과표명개진후적산법시가행적;재선취유전산법여개진산법진행대비,결과표명개진후적산법재심우정학도、시간효솔화은정성방면균우우유전산법.
Population segmentation is an important part of shuffled frog leaping algorithm of which convergence rate is affected by differ-ent population segmentation methods. A new segmentation method on the basis of the original SFLA is proposed in this paper so that the individual fitness in each ethnic group approaches equilibrium. Then present a new learning strategy combined with the deviated path idea of the Yen algorithm to improve the algorithm' s iterative method. The improved shuffled frog leaping algorithm is suitable to solve the K-shortest path problem. Finally,do the simulation for the improved algorithm. Firstly,the monotonous recursive Dijkstra algorithm is se-lected to verify the feasibility of the improved algorithm and the results show that the improved algorithm is feasible. And then the genetic algorithm is selected to compare with the improved algorithm,and the results show that the improved algorithm' s optimization accuracy, time efficiency and stability is superior to the genetic algorithm.