Evaluation of microstructure and prediction of hardness of Al–Cu based composites by using artificial neural network and linear regression through machine learning technique

Al–Cu alloy with B4C particulates will meet the specific application which includes window panel, seats, aircraft structure and aircraft fittings due to their excellent mechanical properties. In this paper, Al–Cu/B4C composites was fabricated by using three parameters (wt% of B4C, ageing duration and mesh size) with three level each as per the design of experiments. Al–Cu/B4C composites were machined as per IS:1500 standard to evaluate hardness by experimental method. An Artificial Neural Network model was developed by using the Levenberg–Marquardt algorithm to predict the experimental hardness value of formed composites. Linear Regression model is created and evaluated by taking 30% of experimental data set for testing and 70% for training. Polynomial feature is imported with only 2° with their interaction only. It is seen that the established ANN model predicts the closeness with the experimental hardness within ± 10% error. It is seen that 14.58% improvement was been observed after considering polynomial feature for the linear regression model. In addition, microstructure study was discussed for the fabricated composites as per IS:7739 standard and observed that B4C particles were homogeneously dispersed in the Al–Cu based matrix and exhibit good bonding between them. © The Author(s), under exclusive licence to Springer Nature Switzerland AG 2024.