NSGA-II-RJG applied to multi-objective optimization of polymeric nanoparticles synthesis with silicone surfactants

Abstract

AbstractPolydimethylsiloxane nanoparticles were obtained by nanoprecipitation, using a siloxane surfactant as stabilizer. Two neural networks and a genetic algorithm were used to optimize this process, by minimizing the particle diameter and the polydispersity, finding in this way the optimum values for surfactant and polymer concentrations, and storage temperature. In order to improve the performance of the non-dominated sorting genetic algorithm, NSGA-II, a genetic operator was introduced in this study — the transposition operator — “real jumping genes”, resulting NSGA-II-RJG. It was implemented in original software and was applied to the multi-objective optimization of the polymeric nanoparticles synthesis with silicone surfactants. The multi-objective function of the algorithm included two fitness functions. One fitness function was calculated with a neural network modelling the variation of the particle diameter on the surfactant concentration, polymer concentration, and storage temperature, and the other was computed by a neural network modelling the dependence of polydispersity index on surfactant and polymer concentrations. The performance of the software program that implemented NSGA-II-RJG was highlighted by comparing it with the software implementation of NSGA-II. The results obtained from simulations showed that NSGA-II-RJG is able to find non-dominated solutions with a greater diversity and a faster convergence time than NSGA-II.