Performance Comparison of Genetic Algorithm Operator Combinations for optimization Problems

Abstract

Genetic Algorithm (GA) is a meta-heuristic search algorithms and the process of GA inspired by natural evolution theory. GA has chromosome-forming operators i.e. selection, crossover, and mutation. These operators have different techniques to solve optimization problems. This paper presents performance comparison of 36 combination operators from 4 selection techniques, 3 crossover techniques, and 3 mutation techniques. Combinations of selection, crossover and mutation operators are simulated for optimizing three benchmark function: Rastrigin function, Sphere function, and Exponential Function. The purpose of this study to find suitable combinations of selection, crossover, and mutation technique for each objective function tested. The performance to be analyzed and compared in this study are average iteration, the best value of objective function f(x)/fitness value, xi value, and average time. The result shows that different selection, mutation and crossover have different effectiveness to optimize various objective functions, and for each objective function have different suitable GA operator combination to find the optimal solution.