Force density-informed neural network for prestress design of tensegrity structures with multiple self-stress modes

Abstract

In recent years, the tensegrity structures have been studied and applied extensively in the engineering field because of their unique topology, shape, and stability. An effective force density-informed neural network (FDINN) approach for performing a robust force finding procedure is constructed in this paper by utilizing a fully connected neural network. The design variables as the input values are first transformed into a parameterized neural network, then these neural network’s parameters are updated in the learning process by an error design loss function via backpropagation and standard optimizer to obtain an optimal self-stress configuration as the output values. By using FDINN, this approach is effectively applied for multiple tensegrity models without depending on the amount of data samples and considering the structural symmetric property. The stability of tensegrity structure is investigated by checking tangent stiffness matrix. Various numerical examples are thoroughly carried out to validate for the effectiveness of the present approach.