Cluster-Based Multiobjective Particle Swarm Optimization and Application for Chemical Plants

In multiobjective particle swarm optimization (MOPSO), the global-best is randomly selected for each population from a nondominated solution set. However, this Roulette wheel-based global selection ineffective convergence and diversity when problem has numerous decision variables or large number of candidates. Thus, study proposes cluster-based MOPSO (CMOPSO). CMOPSO, similarities between particles are considered selecting particle. The cluster determined based on Euclidean distance in objective space. proposed approach demonstrated by applying an operating condition to hydrogen production process. target process representative chemical plant with search space strong nonlinearity. Furthermore, performance CMOPSO assessed comparing it that MOPSO. results indicate exhibits superior terms diversity.