Novel Prediction Model for Validating the Mechanical Behaviour of Composite Materials Using Deep Learning

In the research problem, a deep learning technique is used to predict the mechanical parameter of a hybrid fibre reinforced composite material. To improve the performance of the modal, a Bidirectional Long Short-Term Memory (BiLSTM) model with the Modified African Vulture Optimization Algorithm (MAVOA) is included. To gather a data set for the neural network's training, experimental methods were used. Epoxy resin with sisal and banana fibre is cast in multiple proportions and tested in accordance with its flexural, impact, and tensile strengths. The proportion of epoxy resin, sisal fibre, and banana fibre is used as the model's input parameter, and the output is its mechanical properties. At the phase of exploration and exploitation, the African Vulture Optimization Algorithm is modified using Brownian motion. To compare the suggested modal with the BiLSTM, LSTM, and Gated recurrent units, the Mean Square Error, Mean Absolute Error, and Root Mean Square Error were calculated from the modal R-Square for the predicted value. For training and testing, the dataset was split into 80 and 20% This suggested modal will acquire an approximation of the mechanical property of the predetermined ratio of resin to natural fibre. The proposed model's R-Square achieved about 98%. MSE, MAE, and RMSE values are 0.1, 0.25, and 0.2, respectively.