Artificial neural network-aided force finding of cable dome structures with diverse integral self-stress states-framework and case study

Abstract

Force finding of cable dome structures is a sophisticated and indispensable procedure in terms of their unique structural stiffness derived from self-equilibrium stress to the tension cables and compression struts. Owing to the limitation of traditional force finding methods, the process is often time-consuming and challenging. To address this issue, a general machine learning-aided computational framework for conducting a reliable force finding process is established. Several commonly used force finding methods for cable dome structures are summarized at first. Then the paper introduces the fundamental principle of the artificial neural network (ANN) methods including Back Propagation neural network (BPNN), Radial Basis Function neural network (RBFNN) and General Regression neural network (GRNN), and proposes to combine ANN methods with finite element analysis (FEA) to the force finding of cable dome structures. Additionally, the Geiger, Kiewitt, Levy and hybrid cable dome structures (Tianquan Gymnasium in Sichuan Province, China) are taken as examples of case validation, solved by ANN-aided force finding methods respectively. The results indicate that the proposed computational framework is capable of searching for the feasible prestress whether or not considering the external load such as self-weight. Among ANNs in this case study, GRNN can be applied in the calculation of not only basic cable dome structures with both single and multiple integral self-stress states, but also intricate cable dome structure in practical engineering projects with optimal accuracy and efficiency.