An Investigation and Improvement of the Performance of the GA–DE Hybrid Evolutionary Algorithm

Optimum dimensional synthesis of the five-point double-toggle mould clamping mechanism for thrust saving performance has been successfully solved using a genetic algorithm– differential evolution (GA–DE) method. To further validate the performances of the GA–DE algorithm in terms of its search ability (accuracy), efficiency, reliability and robustness across the widest possible range of functions, a test-suite of 20 functions of 1–20 variables discussed in the literature is performed. The premature convergence, related to the reliability performance, for the optimum dimensional synthesis problem using the GA–DE algorithm has not been investigated and improved in the previous work. Thus, a scheme of combining certain excellent individuals as the disturbed vectors and the technique of a large initial population size is proposed to improve the premature convergence for the GA–DE algorithm. The results obtained by the GA–DE algorithm are compared with those obtained by the other four evolutionary algorithms. Findings show that the GA–DE algorithm generally has very good search ability, efficiency and robustness. Lastly, it can also be seen that the proposed scheme can improve the reliability of the GA–DE algorithm for the test functions and the optimum dimensional synthesis problem.