Particle Swarm Optimization for Multi-response Parameter Optimization Based on Desirability Functions

In order to solve the multi-response parameter optimization problem faced by manufacture industry, particle swarm optimization was employed to find the maxima of the desirability function. By designing the experiment of central composite design, analyzing and testing significance of the polynomial models, constructing desirability function and finding the best factor levels based on the fitness function, we may obtain the model that can solve the optimization problem rationally. This paper employs a numerical example for illustration, and achieved with objective value 0.4377 in iteration less than 10. Empirical results reveal that the PSO algorithm based on desirability functions has powerful global search ability and high convergence on solve the optimization problem.