A topology and sizing optimization method for frame structures combined with an orthogonal-maximin Latin hypercube design(LHD) method

Abstract

Abstract This work presents a new strategy for improving the global searching capability of the two-level approximation method. By using an orthogonal-maximin latin hypercube design (LHD) method, five initial designs reflecting more information of the feasible domain can be generated. Due to their good representation for the feasible domain of the five latin hypercube designs, it is more likely to obtain a global optimum or a near global optimum which is better than before. Numerical examples of frame structures are established to illustrate the effectiveness of this method. From the results, it can be seen that a series of better optimal solutions can be obtained by using the strategy of orthogonal-maximin LHD method and the number of structure analysis required in the optimization process is dozens of times or fewer, even comparable to size-variable optimization.