| As one of the Swarm Intelligence Algorithms, ant colony algorithm is inspired bythe foraging behavior of ant colonies. Since Dorigo M proposed this algorithm in1991, some people have studied it and presented many improvements based on theinitial ant colony algorithm. Genetic Algorithm is a programming technique thatmimics biological evolution as a problem-solving strategy. For the specific problem,GA find a solution through some genetic manipulations: select, intersect and change.And GA can evaluate each candidate of the specific problem by a metric called fitnessfunction. Now the GA and ACO have become two of the most famous optimizationalgorithms, and they are widely used in many fields.These two algorithms have some common characteristics, such as parallelism,universal adaptability and so on. But they also have their own advantages anddisadvantages. GA has good global search ability and it can keep the diversity ofsolutions, but it has no feedback mechanism. ACO can use the historical experiencesto help search the solution, but it will spent a long time and be easy to find the localoptimal solution. So in this paper would like the two algorithms fusion, and find aGACO or GEACO algorithm that can improve them two.In combinatorial optimization problems, the fusion algorithm use ACO to solvethe TSP problem and use GA to update the parameters of ACO. This method canimprove the search results, and when we get the solution of the problem at the sametime also can have a preliminary of the parameters combination.In continuous optimization problems, first for guiding the ACO’s initial positiondistribution, the fusion algorithm uses GA to get the initial pheromone distribution.Then algorithm uses ACO to search the solution of the problem. During the ACO step,it also uses GA manipulations to prevent algorithm falling into the local optimum.According to the two kinds of fusion method, this paper uses matlab to verify thefeasibility of them, and analyzes their performances. |
PDF Full Text Request