A New Nature-Inspired Hybrid Algorithm with a Penalty Method to Solve Constrained Problem

Abstract

In this paper, an efficient hybrid method for solving constrained numerical and engineering optimization problems is presented. The proposed hybrid method uses a combination of three methods a genetic algorithm (GA), particle swarm optimization (PSO), and symbiotic organisms search (SOS) to find desired solution of problems in a complex design space and to control the feasibility of finding solutions using the penalty function method. There are three alternative phases in the proposed algorithm: GA, which develops and selects the best population for the next phases, PSO, which obtains the experiences for each appropriate solution and updates them, the SOS, which performs symbiotic interaction updates in real-world populations. For constraint handling, the penalty function method was used with same values to control the parameters based on the problem. The proposed algorithm was tested on a set of constrained problems from previous studies. Normality and non-parametric tests were performed on the proposed method. the obtained results have showed that the proposed method is able to score the best rank among the CEC 2010 competition algorithms. It is able to solve most of these numerical and engineering design problems up to the best desired solutions and reach the minimum number of function evaluations. The proposed method had a showed better performance than the [Formula: see text]DEag, which was the winner of the CEC 2010 competition.