Pareto optimal design of square cyclone separators using a novel multi-objective optimization algorithm

In the present study, multi-objective optimization (MO) of square cyclones is performed in three steps. In the first step, the collection efficiency (h) and the pressure drop (Delta p) in a set of square cyclone separators are numerically investigated using computational fluid dynamics techniques. In the second step, two meta-models based on the evolved group method of data handling-type neural networks are obtained, for modelling of eta and Delta p with respect to geometrical design variables. Finally, a novel MO based on a combination of the particle swarm optimization, multiple-crossover and mutation operator is introduced. The proposed MO is applied to Pareto optimal design of square cyclones considering two conflicting objectives (eta and Delta p), based on the obtained polynomial neural networks. Furthermore, the proposed Pareto result is compared with that of Non-dominated Sorting Genetic Algorithm II.