Particle swarm optimization algorithm with moving boundaries as a powerful tool for exploration research

An advanced modification of the particle swarm optimization (PSO) algorithm is presented. The proposed modification consists in implementation of floating boundaries for the design space whose positions are moved (shrink or expand) during computations depending on the number of particle hits. As demonstrated by example of three test functions, such an adaptive “moving boundary PSO” algorithm outperforms its classical analog in terms of both convergence and success rates. Moreover it is capable of finding the global extremum even if it is located outside the initial design space.