Improved Bacterial Foraging Algorithm

Bionic intelligent algorithm is a kind of random search algorithm proposed to simulate the biological group behavior of the human being. It has a wider application than the traditional optimization algorithm because of no requirements in the smoothness of target functions, globally searching, parallelism and so on.This paper studies mainly the bacterial foraging algorithm, which is a kind of intell-igent algorithm to simulate the foraging behavior of the human e. coli, and search the solution space randomly using the laws of "evolution and survival of the fittest" in nature. Then it has no need of the gradient information of the functions. But the algorithm contains three-layer nested loops, where the outer is the disperse cycle, the middle is the replication cycle and the inner layer is chemotactic cycle. Thus it has complex structure, slow convergence speed and more parameters, and is not suitable to solve the problems of mass caused in the large-scale case. To overcome these shortcomings, two kinds of improved bacterial foraging algorithms are given in this paper:First, a bacterial foraging algorithm with adaptive search step is proposed in this paper. In the disperse operations new individuals are generated by using the Gaussian distribution around the best individual to reduce the possibility of elite individuals to be eliminated, and improve the convergence speed of the algorithm. Finally the effectiveness of the proposed algorithm are illustrated by the standard test functions of numerical experiments.Second, to strengthen the purpose of the algorithms in searching, and to improve the convergence speed and accuracy, a simulated bacterial foraging algorithm with direction is presented based on the adaptive bacterial foraging algorithm. New direction is produced by combination of a approximating gradient and the random direction with a weight. The effectiveness of the algorithm was illustrated by several typical testing functions, and the estimate redults show that its optimization ability is superior to the adaptive simulated bacterial foraging algorithm.