GHOST—Gate to Hybrid Optimization of Structural Topologies

摘要

Although well-recognized in the fields of structural and material design and widely present in engineering literature, topology optimization still arouses a high interest within research communities. Moreover, it is observed that the development of innovative, efficient and versatile methods is one of the most important issues stimulating progress within the topology optimization area. Following this activity, in the present study, a concept of a hybrid algorithm developed in order to generate optimal structural topologies of minimal compliance is presented. The hybrid algorithm is built based on two existing approaches. The first one makes use of the formal optimality criterion, whereas the second one utilizes a special heuristic rule of design variables updating. The main idea that stands behind the concept of the present proposal is to take the advantage of both algorithms capabilities. In a numerical implementation of the hybrid algorithm, the design variables are updated at each iteration step using both approaches, and the solution with a lower objective function value is selected for the next iteration. The numerical tests of the generation of minimal compliance structures have been performed for chosen structures including a real engineering one. It has been confirmed that the proposed hybrid technique based on switching between the considered rules allows the final structures having lower values of compliance as compared with the results of an application of basic algorithms running separately to be obtained. Moreover, based on promising results of the tests performed, one can consider the proposed concept of a hybrid algorithm as an alternative for other existing topology generators.