Diversity control in particle swarm optimization

Population diversity of particle swarm optimization (PSO) is important when measuring and dynamically adjusting algorithm's ability of “exploration” or “exploitation”. Population diversities of PSO based on L
₁
norm are given in this paper. Useful information on search process of an optimization algorithm could be obtained by using this measurement. Properties of PSO diversity based on L
₁
norm are discussed. Several methods for diversity control are tested on benchmark functions, and the method based on current position and average of current velocities has the best performance. This method could control the PSO diversity effectively and gets better performance than the standard PSO.