A regional genetic algorithm for the discrete optimal design of truss structures

A regional genetic algorithm (R-GA) is used for the discrete optimal design of truss structures. The chro- mosomes are selected from a sub-region centred on the continuous optimum. This approach replaces genetic rebirth as previously proposed by other authors, thereby signi cantly reducing computational costs. As a pure discrete method, the R-GA method does not require heuristic arguments or approximations. This makes the algorithm highly e ective when buckling and slenderness constraints with scatter in the data are introduced. A large set of numerical test examples is used to illustrate the capabilities of the method. The algorithm is shown to be e ective and robust, making it suitable for the optimal design of very large truss structures. Copyright