Artificial Bee Colony Algorithm And Its Application

Traditional optimization algorithms are not suitable for solving large-scale problems with high dimension and nonlinear or other special properties. Swarm intelligence algorithm to make up for the defects of traditional optimization algorithm. They made a better effect in the application of many difficult problems. Artificial bee colony algorithm is a new swarm intelligence algorithm, it has many advantages, but there are discarded solutions and it is easy to fall into local extremum, so it has great significance to the improvement of algorithm. The application of the algorithm is a hot research point. This paper has made the improvement to the artificial bee colony algorithm, the improved algorithm was used to solve the function optimization problem, TSP problem and UCI data clustering problems, results of these problems were analyzed and parameters analysis. Finally, the parameter and results analysis show that the performance of the modified algorithm is better compared with the traditional clustering algorithm.The research work and innovation points of this paper are as follows:1、Artificial bee colony algorithm was introduced briefly. The Markov chain model was used to prove that the artificial bee colony algorithm with finite homogeneous Markov chain model, which proved that the bee colony algorithm is convergent.2、The research on artificial bee colony algorithm has often only focus on one aspect of defect, this paper had improved the algorithm of early and late period. Negative feedback mechanism was introduced in the early period, the way that eliminate bad fitness nectar directly previously was changed, the bad nectar was turned to the scouting bee.Fitness worse the search step length is longer which formate of "jump" search. This reduces the chance of losing the optimal solution, increasing the possibility of solution space full of space exploration. Cross immune mechanism is introduced at the end of the algorithm, two good nectar source was selected to make crossover operation, so that the algorithm is easy to jump out of local optimal value. New nectar after crossing due to excellent senior information, thus have a greater possibility is good nectar source, and it will not cause the algorithm does not converge.3、The improved algorithm was applied to the function optimization problem, and the parameters and results was analyzed, through comparative analysis shows that bee colony algorithm had good performance.4、The artificial bee colony algorithm was applied to the traveling salesman problem, and the results was analyzed, through the analysis and comparison to the result of particle swarm optimization, proving artificial bee colony algorithm has a better performance.5、The bee colony algorithm was applied to UCI data clustering problems, and the parameters and results was analyzed, through comparative analysis shows that ABC algorithm can solve the clustering problem and has a good effect.