An improved genetic algorithm with initial population strategy and self-adaptive member grouping

The performance of genetic algorithms (GA) is affected by various factors such as coefficients and constants, genetic operators, parameters and some strategies. Member grouping and initial population strategies are also examples of factors. While the member grouping strategy is adopted to reduce the size of the problem, the initial population strategy is applied to reduce the number of search to reach the optimum design in the solution space. In this study, two new self-adaptive member grouping strategies, and a new strategy to set the initial population are discussed. Previously proposed self-adaptive approaches for both the penalty function and the mutation and crossover operators are also adopted in the design. The effect of the proposed strategies on the performance of the GA for capturing the global optimum is tested on the optimization of 2d and 3d truss structures. It is worthy to say that the proposed strategies reduce the number of searches within the solution space and enhance the convergence capability and the performance of the GA.