Data-driven failure prediction of Fiber-Reinforced Polymer composite materials

Abstract

The present study illustrates the effectiveness of Deep Neural Networks (DNN) as a tool for creating a data-driven failure model for Fiber-reinforced Polymer (FRP) composite materials. Experimental failure data presented in the literature for laminates tested under biaxial and triaxial stresses were used to develop the data-driven model. A fully connected DNN with 20 input units and 1 output unit trained with a constant learning rate. The network’s inputs describe the laminate layup sequence, lamina properties, and the loading conditions applied to the test specimen, whereas the output is the length of the failure vector. The failure boundaries generated by the DNN were compared to conventional theories such as the Tsai–Wu, Cuntze, and Pinho theory. The data-driven model’s predictions are found to fit the experimental data better than the conventional theories. The DNN’s ability to fit higher-order polynomials to data makes it an effective tool for predicting the final failure of FRP composite laminates .