Design of space trusses using modified teaching-learning based optimization


A modified teaching-learning-based optimization (TLBO) algorithm is applied to fixed geometry space trusses with discrete and continuous design variables. Designs generated by the modified TLBO algorithm are compared with other popular evolutionary optimization methods. In all cases, the objective function is the total weight of the structure subjected to strength and displacement limitations. Designs are evaluated for fitness based on their penalized structural weight, which represents the actual truss weight and the degree to which the design constraints are violated. TLBO is conceptually modeled on the two types of pedagogy within a classroom: class-level learning from a teacher and individual learning between students. TLBO uses a relatively simple algorithm with no intrinsic parameters controlling its performance and can easily handle a mixture of both continuous and discrete design variables. Without introducing any additional algorithmic parameters, the modified TLBO algorithm uses a fitness-based weighted mean in the teaching phase and a refined student updating process. The computational performance of TLBO designs for several benchmark space truss structures is presented and compared with classical and evolutionary optimization methods. Optimization results indicate that the modified TLBO algorithm can generate improved designs when compared to other population-based techniques and in some cases improve the overall computational efficiency. © 2014 Elsevier Ltd.

Publication Title

Engineering Structures