Prediction of tensile properties in 3D printed GF/RPP composites using MISCSO-BP neural network

This study proposes a backpropagation (BP) neural network model based on the multi-strategy improved sand cat swarm optimization (MISCSO) algorithm to predict the tensile properties of glass fiber-reinforced recycled polypropylene (GF/RPP) composites under different fused deposition modeling (FDM) parameter combinations. First, the MISCSO algorithm is built upon the SCSO algorithm by introducing a cubic chaotic reverse learning strategy to enhance population diversity, incorporating the dynamic nonlinear sensitivity range and Weibull flight strategy to strengthen global search capability, and further employing the Gaussian-Cauchy mutation strategy to avoid local optima and accelerate convergence. Subsequently, using printing temperature, layer thickness, infill density, and raster angle as input variables, 243 experimental samples were employed for model training and validation. Model performance was evaluated based on the mean absolute error (MAE), mean square error (MSE), root mean square error (RMSE), and coefficient of determination (R 2 ). Finally, the results show that the proposed MISCSO-BP model achieved R 2 values of 0.93, 0.91, and 0.87 for tensile strength, elastic modulus, and elongation at break, with average prediction errors below 5%, outperforming existing methods in prediction accuracy, convergence speed, and stability. These findings demonstrate the effectiveness and robustness of the MISCSO-BP model in optimizing FDM process parameters and predicting the mechanical properties of sustainable polymer composites.