Proposing a Model to Predict the Bearing Capacity of Hammered Piles Using Genetic and Lorenberg Algorithms

This research aimed to predict the bearing capacity of hammer piles using genetic and Lorenberg algorithms and Machine Learning (ML) methods. The studied samples as input parameters in this research include the parameters of soil internal friction angle, soil elastic modulus, pile diameter (D), and pile length (L) as input to the considered models, and the target in this research is the bearing capacity of the pile. 15% of the input data were considered training data, 15% validation data, and neural network training was done. At first, using the trial and error method, the number of hidden layers was determined as 6, and the target network was trained using the genetic algorithm. The results of training the target network using the genetic algorithm showed that the regression coefficient obtained from the model prediction for the learning and validation data was 99. 0 has been obtained. The results of neural network training using Lorenberg's algorithm showed that the correlation coefficient between training and validation data is 0.96548 and 0.993889, respectively. By comparing the results of the neural network with the laboratory data, it has been observed that the genetic algorithm can make the desired prediction better.