An improved GBESO method and application for engineering structures

• Published in: Structures (Oxford) Vol. 57; p. 105083
• Main Authors: Zhang, Huzhi, Liu, Xin, Fang, Zilin, Yin, Bin
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.11.2023
• Subjects: Engineering structure; Optimum design; strut-and-tie model (STM); the Bi-directional Evolutionary Structural Optimization (BESO); the Genetic Bi-directional Evolutionary Structural Optimization (GBESO); Topology Optimization; Engineering structure; Optimum design; the Genetic Bi-directional Evolutionary Structural Optimization (GBESO); the Bi-directional Evolutionary Structural Optimization (BESO); strut-and-tie model (STM); Topology Optimization
• ISSN: 2352-0124; 2352-0124
• DOI: 10.1016/j.istruc.2023.105083

For engineering structures, the evolutionary-type optimizations are usually conducted to obtain the topologies by deleting and restoring a few of their elements in each iteration. As for the Genetic Bi-directional Evolutionary Structural Optimization (GBESO) among them, five major improvements are made in this research: regroup only the remained elements into three groups in each iteration; perform evolution and punishments on the specific elements during each iteration; use a constant probability throughout the crossover; add a global mutation with a very small probability, and adjust half of the genetic codes of the deleted elements from 0 to 1 as the restoration, whose probabilities are determined by the sensitivity of the elements around them. These improvements are intended to increase the participation of probabilistic ideology, and then to reduce the possibility of obtaining the local optimal solutions. As a result, within a large scope all the time, the improved GBESO keeps searching for global optimal solutions more consistent with the optimization objective and helps establish superior strut-and-tie models (STMs) with a reasonable process, indicating its stronger ability on global optimization.