Usage of neural networks for analysis and processing of experimental research of composite materials

Modern industrial development demands the creation of new materials to enhance the durability and operational lifespan of machines while reducing metal and energy consumption. Composite materials, particularly those based on polymers (reactoplasts), play a key role in achieving these goals. Neural networks, including CNNs, RNNs, LSTMs, GANs, and transformers, outperform traditional algorithms in pattern recognition tasks and are effective tools for analyzing macro- and microstructures of composite materials with predefined properties. Despite challenges in training deep models, requiring significant computational resources and energy, optimization methods like quantization and distillation help reduce costs. Integrating quantum computing further accelerates optimization processes. The importance of Explainable AI is emphasized to address the «black-box» nature of neural networks, ensuring their reliability for critical applications. These technologies are essential for advancing intelligent systems and creating next-generation materials for high-tech industries.