Optimal genotypic feedback in genetic algorithm : a new optimization approach

Genetic algorithm (GA) is a promising means to solve engineering optimization problems. GA is able to perform the global search with minimal simplifying assumptions about the problem as well as the correcsponding decision space. GA faces problems like premature convergence and slow convergence due to decreasing population diversity. To surmount this problem, we have developed a new approach of optimal genotypic feedback (OGF). This approach generates binary building blocks of random size from the optimal solution after each generation.The blocks are then inserted in the subsequent generation. This new approach is successfully tested on number of nonlinear, multimodal and non-continuous optimization problems. The results demonstrate that the approach efficiently searches good quality solutions.

In the next step, OGF is amagamated with hybrid GA (HGA). The resulting new HGA is applied on six optimization problems involving characterization parameters of pulp chest and miminum variance control. Comparisons with the old HGA indicate the equivalence of OGF with gradient search. Furthermore, the new HGA is observed to yield results in less number of objective function evaluations.