Real-coded genetic algorithm-based particle swarm optimization method for solving unconstrained optimization problems

In solving unconstrained global optimization (UGO) problems, devising nonlinear programming (NLP) methods based on gradient information are extremely difficult when an objective function is non-differential. As a stochastic global optimization algorithm, particle swarm optimization (PSO) algorithm does not require gradient information, enabling it to overcome the limitation of traditional NLP schemes. Unfortunately, performance of a PSO algorithm depends on several parameters, such as constriction coefficient, cognitive parameter and social parameter. To overcome the above limitations of a PSO algorithm, this work presents a real-coded genetic algorithm (RGA)-based PSO (RGA-PSO) algorithm. The specific parameters of the inner PSO algorithm are optimized using the outer RGA. Performance of the proposed RGA-PSO algorithm is then evaluated using a set of UGO problems. Numerical results indicate in addition to its ability to converge to a global minimum for each test UGO problem, the proposed RGA-PSO algorithm provides a solution is more precise than those of some stochastic global optimization algorithms. Thus, the RGA-PSO algorithm can be considered as an alternative stochastic global optimization scheme for solving UGO problems.