Evaluation of Neural Network Model for Estimating Pile Load Capacity

Neural network models based on deep learning algorithms are increasingly used for estimating pile load capacities as supplements of bearing capacity equations and field tests. A series hyperparameter tuning is required to improve the performance reliability developing a neural model. In this study, number hidden layers neurons, activation functions, optimizing gradient descent method, rates were tuned. The grid search method was applied tuning, which hyperpameter optimizer supplied by platform. cross-validation enhance model validation. An appropriate epochs determined using early stopping prevent overfitting training data. tuned optimum evaluated test data set revealed that could estimate approximately with an average absolute error 3,000 kN coefficient determinant 0.5.