INTELLIGENT SYSTEM FOR THE AUTOMATION OF COMPOSITE MATERIAL CHARACTERISTICS RESEARCH USING NEURAL NETWORKS

Abstract

The development of the scientific and technological sector of Ukrainian industry in the design of mechanisms and machines with improved performance and reduced material and energy consumption necessitates the use of composite materials and coatings based on them. The creation of new polymer-based composites is a promising approach to addressing this challenge. The most effective solution is the development of materials based on an epoxy matrix. This makes it possible to produce epoxy composite coatings on large-scale surfaces with complex geometries.

The development of such materials involves the use of modifiers and fillers of various origins to create multifunctional coatings. During experimental investigations, a significant challenge arises in processing large volumes of experimental data. Traditional approaches based on extensive experimental studies followed by data analysis require considerable time and financial resources. The automation of experimental data processing involves the application of neural network technologies for predicting the properties of composites both at the stage of material development and during their service life.

A comprehensive scientific approach based on the integration of mathematical modeling methods and advanced artificial intelligence architectures enables the planning and processing of experimental data for the construction of highly accurate and interpretable models using deep neural networks. The implementation of the proposed approach will facilitate the creation of self-learning research systems capable of designing composite materials with predetermined properties for modern industrial applications.