AbstractThis study develops an intelligent non-dominated sorting genetic algorithm (GA), called INSGA herein, which includes a non-dominated sorting, crowded distance sorting, binary tournament selection, intelligent crossover and non-uniform mutation operators, for solving multi-objective optimization problems (MOOPs). This work adopts Goldberg's notion of non-dominated sorting and Deb's crowded distance sorting in the proposed MOGA to achieve solutions with good diversity-preservation and uniform spread on the approximated Pareto front. In addition, the chromosomes of offspring are generated based on an intelligent crossover operator using a fractional factorial design to select good genes from parents intelligently and achieve the goals of fast convergence and high numerical accuracy. To further improve the fine turning capabilities of the presented MOGA, a non-uniform mutation operator is also applied. A typical mutation approach is to create a random number and then add it to corresponding original value. Performance evaluation of the INSGA is examined by applying it to a variety of unconstrained and constrained multi-objective optimization functions. Moreover, two engineering design problems, which include a two-bar truss design and a welded beam design, are studied by the proposed INSGA. Results include the estimated Pareto-optimal front of non-dominated solutions.
Genetic Algorithm for the Multi-Objective Optimization of Product Gene
