The use of neural network models is currently a trend. These models are used in a number of fields, particularly in practice-oriented scientific research. Neural network models allow us to predict results and determine the effectiveness of potential solutions.

Our research team uses these models for the complex processing of polymer composite materials (PCMs) produced using additive manufacturing technologies and consisting of continuous carbon fiber and a superstructural polymer, polyetheretherketone. The processing is performed in a microwave field. Applying this processing to fully fabricated PCMs improves their physical and mechanical properties. This is achieved through a number of simultaneous processes, including heating the sample, intense heating at the matrix-fiber interface, localized melting of the polymer, and others. The complexity of the experimental work precludes the collection of a large amount of empirical data. Therefore, we use a series of models interconnected in a specific manner. We train them using available data and validate them using one of the quality metrics: MSE, NRMSE, and R². The coefficient of determination (R²) characterizes the proportion of variance in the dependent variable explained by the model.

The analysis of a series of models yielded R² results ranging from 0.87 to 0.99. These values characterize the high quality of the resulting models and indicate the absence of overfitting.

The work was supported by the Russian Science Foundation grant 23-79-00039.