This paper describes research aimed at the experimental determination of the influence of rubber hysteresis on the friction coefficient between rubber samples for making soles and granite tiles. In the experiments, four types of shoe rubber with similar hardness and different hysteresis properties and two granite tiles with different roughnesses (smooth and anti-slip) were used. The determination of rubber hysteresis was carried out experimentally on a uniaxial testing machine. The friction coefficient was measured using a device specially developed for this type of test, which was based on the pulling force method, while the measurement conditions were based on the EN 13893:2011 standard. The friction coefficient was measured at two different speeds, 50 mm/s and 300 mm/s, with different surface conditions. Using regression analysis and the Taguchi method, the data obtained from the experiments were analyzed to determine the influence of parameters on the friction coefficient. The experimental research shows that different rubber mixtures with the same or similar hardness could have different hysteresis properties but also different friction properties.

This paper present research focused on the prediction of the friction coefficient of shoe sole rubber by utilizing its measured hysteresis values, along with other influencing factors such as hardness, tile surface roughness, sliding speed, and surface conditions. Previous authors research determined that rubber hysteresis is an important property of rubber (among other mechanical and physical properties) to consider when performing tribological research of contact between rubber soles and a hard substrate (tiles, laminate, vinyl, concrete). Data required for design and training of a neural network were gathered by friction coefficient testing conducted on a specially designed test apparatus. Additionally, rubber hysteresis data were obtained using a uniaxial tensile testing machine. Given the role of rubber hysteresis in determining its properties, this study identifies it as a parameter that influences the friction coefficient and aids friction coefficient prediction through artificial neural networks (ANN). The research results showed a high correlation between the friction coefficient values predicted by ANN and actual experimental results, confirming that designed ANN can be used to predict the values of friction coefficient when the rubber hysteresis value is known.