Ordered Replacement-A Different Genetic Selection Algorithm

Searching the optimal solution for constrained optimization problems take more computational time in locating feasible solutions, when the percentage of feasible region within the search space is very small. Genetic algorithms prove its robustness in finding the optimal solution and necessitate more population diversity in locating the feasible region. The selective pressure has to be increased in such a way that the search does not converge to local optimal solution. Mixed sampling mechanism is opted to accomplish these conditions, which involves both stochastic and deterministic selection procedures. The algorithm presented contains these features and is attained by improving the fitness of solutions within a generation before the next generation is produced. Three test problems are solved, with the proposed ordered replacement genetic algorithms to improve the fitness of the population in every generation. Further generations are produced by proportionate selection method or tournament selection method. Both real and binary coded genetic algorithms are used to solve the test problems. An approach for optimal design of final drive in passenger bus using the proposed algorithms is presented with the objective to minimize the volume. The design search space is defined by the decision variables such as transverse module, face width, number of pinion teeth, number of gear teeth and is further influenced by the constraints such as bending stress, contact stress, face contact ratio and transmission ratio. The optimal design problem is solved using the proposed method. The results show the proficiency of ordered replacement genetic algorithms.