Optimal design of main girder of large pressing machine based on father-offspring combined selection GA

Main girder is one of the most important forced parts of large pressing machine. Its optimal design is a complex real world engineering optimization problem with a series of design constraints. Classical optimization algorithms such as the downhill simplex and sequential quadratic programming are generally not successful in solving such hard optimization problems. They often do not give "global" or "near-global" optimum solutions. In this paper, a modified genetic algorithm is presented which is based on father-offspring combined selection strategy. In the proposed algorithm, a core technique, father-offspring combined selection strategy, is used to improve the convergence drawbacks of the basic genetic algorithm in search for global optimum. The efficiency of the proposed genetic algorithm is illustrated by application to main girder optimum design. The results show that the global searching ability and the convergence speed of this proposed algorithm are significantly better than traditional optimization methods. Hence, the father-offspring combined selection genetic algorithm is efficient and valid in solving practical engineering design problems.