ANALYSIS OF BENEFITS OF INTEGRATING THE OPPOSITION BASED LEARNING TECHNIQUE INTO NON-DOMINATED SORTING GENETIC ALGORITHM III

Abstract

Deterministic and heuristic approaches have shown significant contributions in solving the mono-objective optimization problems. In the case of multi-objective optimization problems, metaheuristic approaches have achieved great success in providing quality Pareto-optimal solutions, rather than deterministic and heuristic approaches. Basically, quality of Pareto-optimal solutions depends upon convergence and diversity maintained by optimization algorithm, where convergence and diversity refer to the search capabilities of optimization algorithm towards and along Pareto-optimal front, respectively. In this line, the non-dominated sorting genetic algorithm III (NSGA III), a population-based metaheuristic method, has become a widely accepted multi-objective optimization algorithm. However, random generation of initial population might generate the undiversified initial population and that may lead to premature convergence giving the local Pareto-optimal solutions rather than global Pareto-optimal solutions. To alleviate, this paper integrates the opposition-based learning (OBL) method in NSGA III for population initialization and generation jumping. The usefulness of OBL integrated NSGA III is demonstrated through solving a numerical example of construction project. Based on several performance metrics, the results of numerical example indicate that the OBL integrated NSGA III outperforms the NSGA III and other existing algorithms in a significant manner.

Received: October 20, 2022;
Revised: November 23, 2022;
Accepted: December 28, 2022;