Development of a hybrid technique for continuous/discrete optimization

Many engineering problems are associated with both continuous and discrete design parameters. While continuous optimization procedures are well established and discrete techniques are becoming more widely used, there is a need for the development of methods which can appropriately include both types of variables. This paper presents a hybrid optimization technique which incorporates continuous and discrete design variables simultaneously. A simulated annealing algorithm is used in conjunction with a gradient based search in the formulation. Numerical examples are presented to demonstrate the procedure. In the first, the mass of a ten bar truss is minimi/,ed using continuous parameters and discrete variables which are allowed to vary from within a set of preselected values. Results indicate that the hybrid optimization procedure produces a better optimum in fewer iterations compared to other existing techniques when applied to a ten bar truss mass minimization. An application to maximize the buckling load of a composite cylinder is also presented.