The use of artificial neural networks in creating ceramic and refractory materials

Using artificial neural networks (ANNs) in chemical and technological calculations to create ceramic and refractory materials allows for high accuracy in predicting the mechanical properties of final products, including cold crushing strength. This significantly reduces experimental costs, as the model can generate predictions for various compositions and process parameters without the need for many physical tests. The paper considers the use of ANNs to predict the mechanical strength of carbon­containing ceramic materials and determine the rational amount of oxidizing agents that intensify carbon burning from ceramic masses containing 20 to 30 wt. % of coal preparation waste. It has been established that the ANNs can provide a comprehensive assessment of the effect of oxidants on the cold crushing strength of the material. Particularly effective are the nitrate mixtures of 80 % wt. % NH4NO3 + 20 % wt. % NaNO3 and 80 % wt. % NH4NO3 + 20 % wt. % Ca(NO3)2, which maximally covers the temperature range and contributes to the maximum increase in cold crushing strength at minimal costs compared to mono­oxidants. The original author’s software “CeramMath” for testing and working with ANNs adapted to the tasks of silicate materials science was created. It has been shown that ANNs are a powerful tool for improving technological processes in producing ceramic and refractory materials, which allows for minimizing experimental costs, improving product quality, and ensuring high production efficiency.