Research On The Combinational Optimization In Shuffled Frog Leaping Algorithm

Swarm intelligence algorithm is an efficient optimization algorithm. Large amount of simple individuals have more advantages over a single complex one. The simple individuals often show better performance on robustness,flexibility and a smaller economic cost. Swarm intelligence provides a new way to solve the combinational optimization problems without global control and global model.The Shuffled Frog Leaping algorithm(SFLA) is a kind of rising swarm intelligence optimizer. Currently, there is rare research on the use of SFLA to solve the combinatorial optimization problems. The thesis presents how to solve the combinatorial optimization problems with the SFLA. The primary contents and results are followingFirstly, the traditional SFLA depends on the global best result and the local best result to improve the worst ones in each sub-population. The strategy imposes on the worst frog in each sub-population. In chapter 3, a new improved strategy was proposed in the simple neighborhood search. The global best result guides each sub-population to achieve a better performance. The strategy is different from the traditional one, it can act on the sub-population instead of a single one.With the city number increases, the traditional evolutionary algorithms may easily fall into the trap of local search, they can not find the global optimal solution. The improved algorithm can efficiently solve the TSP, it can find a satisfactory solution at a small cost.Secondly, the traditional SFLA uses the global best frog and the local best frog to attract the worst frog in each sub-population. They guide the worst one to be a better one. A new improved strategy was proposed in the simple neighborhood search which makes use of the repulsion between the other frogs in the sub-population and the worst frog.In chapter 4, we see the algorithm can efficiently find the KP's global optimal solution.