Prediction of the Thickness of a Boroaluminized Layer Using an Artificial Neural Network

The application of mathematical models and artificial neural networks for predicting the properties of diffusion coatings created by thermal–chemical treatment based on the boroaluminizing process is considered. The formalization and analysis of forecasting experimental results are conducted. Building computer models for prediction based on experimental data of the boroaluminizing process with high accuracy is a solvable task when using artificial neural networks such as a multilayer perceptron. Testing the number of hidden layers and the number of neurons in them revealed the highest correlation coefficient R = 0.99993 for an artificial neural network using two hidden layers with ten and six neurons, respectively. The highest efficiency can be achieved using the hyperbolic tangent activation function.