An improved Bacterial Foraging Optimization algorithm using novel chemotaxis and swarming strategy

摘要

This paper proposes an improved Bacterial foraging optimization (BFO) algorithm based on modified chemotaxis process and novel swarming strategy. Three improvements are presented in the proposed algorithm. First, during the chemotactic process, each bacterium selects one dimension for tumbling randomly to reduce the mutual interference among different dimensions. Second, each bacterium’s movement-length on the selected dimension is determined by the stochastic flight lengths of the improved Levy flight which is composed of many short step-size combined with rarer longer step-size, with this pattern repeated across all scales; moreover, to achieve a better search result, the stochastic step-size is also reduced adaptively based on the evolutionary generations. Third, inspired by the social information term in particle swarm optimization (PSO), the global best solution is used to guild the swarming direction to increase the swarming performance. Experiments are carried out with the aim of studying the performance of the proposed LPBFO algorithm on six widely used functions. The experimental results and analysis demonstrate that the method obtains a marked improvement compared with other competitive algorithms.