Proposing an Improved Genetic Algorithm for Solving Optimization Problems

The genetic algorithm (GA) sometimes fails to achieve an optimum solution since it faces with the problems such as premature convergence and falling into a local optimum. The main reason is that it usually produces the new popualtion through recombining only the promising individuals, and this causes the diversity of population to reduce quickly. To avoid this problem, in this paper, we propose an improved GA (also called ImpGA) in which the unpromising individuals along with the promising ones are participated in the production of the new generation. In addition to selection, ImpGA modifies two main operators of GA, i.e. crossover and mutation. To verify and compare the effectiveness of ImpGA and five other optimization algorithms, namely Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Symbiotic Organisms Search (SOS), Equilibrium Optimizer (EO), and Black Widow Optimization Algorithm (BWO), they are applied to solve ۲۶ unimodal, multimodal, and composition test functions. Experimental results confirm the higher effectiveness of ImpGA compared to the others in terms of obtaining an optimum solution and faster convergence speed

Keywords:

Genetic algorithm , Optimization , Test function , Convergence speed