Optimization of Neural Network architecture and derivation of closed-form equation to predict ultimate load of functionally graded material plate

Functionally Graded Material (FGM) plate is a complicated structure with complex allocation of spatially changing proportions ceramic and metal within the matter. Various analytical numerical methods have been applied view to evaluating critical load FGM plate. However, these conventional struggle when computational complexity significant, which represents an obstacle incorporation other advanced techniques where power required (e.g. optimization or random simulations). The Neural Network (NNet) model has successfully resolve this issue. NNet requires proper configuration take advantage model, thus, careful parameter tuning required. Furthermore, typically “black box,” prediction mechanism hidden. This paper establishes optimized architecture for NNet, parametric study model’s hyperparameters. Variance propagation also observe variation performance on sub-databases splintered from database. To end, explicit expression trained provided after mathematically deploying hidden algebra behind prediction. developed very promising evaluation metrics: R 2 , MAE, RMSE test set are 0.999925, 0.067516, 0.146438, respectively.