Modeling and optimization of the crushing behavior and energy absorption of plain weave composite hexagonal quadruple ring systems using artificial neural network

Abstract Composite structures are being increasingly utilized in the automotive industry for their lightweight and specific energy absorption capabilities. Recently, the Composite Hexagonal Quadruple Ring System (CHQRS) has been proposed as a passive energy absorber device, due to its performance quality. Experiments show that the energy absorption capacity of this system significantly varies dependent on the interior hexagonal ring angle value. Thus, there is a need to develop an accurate model, which can be evaluated based on its performance, to then be used for predictions of its nonlinear loading capacity at various angle configurations and to optimize the interior hexagonal angle for a maximum system energy absorption capability. Within this paper, Artificial Neural Network (ANN) based models have been developed, optimized and compared using the Mean Squared Error (MSE) performance function, with the objective to precisely predict the high non-linear crushing behavior of the CHQRS at different angle configurations, as well as to optimize its internal hexagonal angle for a maximum energy absorption capability. The predicted results and the experimental results have also been compared in terms of their load carrying capacity and energy absorption capability. The developed ANN-based models accurately predict the load carrying capacity and the energy absorption capability at different angles, with an MSE of as low as 0.39 N and 1 J, respectively. The best ANN model has been identified and then utilized to optimize the internal angle of the CHQRS, in order to maximize its energy absorption capabilities; this was found to be 47°, with an energy absorption value of 69 J. For an angle of 35°, the energy absorption was 62 J.