Discrete and continuous variable structural optimization using tabu search

A memory-based combinatorial optimization technique is adapted into a search procedure for solving single and multiobjective structural optimization problems. This approach, known as tabu search, employs two complementary mechanisms, namely tabu restrictions and aspiration criteria for constraining and guiding the search in the design space. A non-Markovian function utilizing recent search history is used to constrain the search by classifying certain moves as forbidden or tabu whereas a short term memory function, providing strategic forgetting, is used to free up the search process. Numerical results obtained using three different tabu search strategies for single and multiobjective design of structures with discrete-continuous variables are presented. To model the multiple objective functions in the problem formulation, a cooperative game theoretic approach is used. The results indicate that, in several instances, the optimum solutions obtained using tabu search outperform the optimum solutions obtained using some gradient-based and discrete optimization techniques. The results also indicate that tabu search has significant potential for additional applications in design optimization, especially for problems with mixed discrete-continuous variables.