Crashworthiness optimization of composite hexagonal ring system using random forest classification and artificial neural network

This research aims to enhance the safety level and crash resiliency of targeted woven roving glass/epoxy composite material for various industry 4.0 applications. Advanced machine learning algorithms are used in this study to figure out the complicated relationship between the crashworthiness parameters of the hexagonal composite ring specimens under lateral compressive, energy absorption, and failure modes. These algorithms include random forest (RF) classification and artificial neural networks (ANN). The ultimate target is to develop a robust multi-modal machine learning method to predict the optimum geometry (i.e., hexagonal ring angle) and suitable in-plane crushing arrangements of the hexagonal ring system for targeted crashworthiness parameters. The results demonstrate that the suggested RF-ANN-based technique can predict the optimal composite design with high accuracy (precision, recall, and f1-score for test and train dataset were 1). Furthermore, the confusion matrix validates the random forest classification model's accuracy. At the same time, the mean square error value serves as the loss function for the ANN model (i.e., the loss function values were 2.84 × 10⁻⁷ and 6.40 × 10^−7, respectively, for X1 and X2 loading conditions at 45° angle). Furthermore, the developed models can predict crashworthiness parameters for any hexagonal ring angle within the range of the trained dataset, requiring no additional experimental effort.