Introduction. Determination of mechanical properties of layered structures of highways is an urgent task. This is due, firstly, to the need to control the quality of new sections during the construction of highways. Secondly, to assess the condition of existing roads with the accumulation of damage and defects is of interest. The formation of multiple defects (cracks) changes the averaged viscoelastic properties of the components of the structure, specifically, the surface asphalt-concrete layers. The article discusses the use of neural network technologies to improve the accuracy of the recovery of viscoelastic properties. This approach is based on experimental methods. As an example, we can give the definition of the dynamic deflection of a structure from a falling weight, FWD.Materials and Methods. The elastic modulus of a three-layer structure was determined on the basis of a neural network. To find out the solution accuracy, it was compared to the results of mathematical modeling and experimental data.Results. The experimental and calculated parameters of the elastic modulus of individual layers of the road structure turned out to be very close. The proposed approach to determining the mechanical properties of materials of road structures allowed us to apply the obtained results to examination of the condition of individual elements and the entire road structure.Discussion and Conclusions. The prospects of using artificial intelligence to determine the mechanical properties of layered structures was shown. Further improvement of methods and tools for analyzing the behavior of road structures under dynamic loading will expand existing approaches to assessing the condition of road structures.

The article is devoted to the development of machine learning methods for classes of technical problems, including determining the properties of materials. According to the authors, the neural network approximation algorithm is able to take into account the behavior of materials in various experimental conditions. The article provides illustrative examples of how a neural network with a single hidden layer can approximate a function of several variables with a given accuracy. As part of the study, a number of experimental measurements were made. The structure of the neural network and its main components are described.

The article sets the task of classifying various materials and determining their belonging to a specified group using a recurrent neural network. The practical significance of the article is to obtain the results of the neural network, confirming the possibility of classifying materials by the hardness parameter using a neural network. As part of the study, a number of experimental measurements were carried out. The structure of the neural network and its main components are described. The statistical parameters of the experimental data are estimated.

No abstract