Indentation is a universal and practical method for obtaining material characteristics, especially when it is impossible or difficult to expose the material to other measuring methods. Experimental data on the mechanical properties of various types of materials were obtained using the shock loading unit. A mathematical model based on the finite element method was used to verify the experimental results. The article considers the solution of the problem of classification of neural metals with different mechanical properties. As part of the work, an artificial neural network has been created that allows the distribution of materials into selected groups. It is determined that a significant advantage of using neural networks is the ability to process experimental data and identify complex nonlinear dependencies, which makes them in demand in tasks related to the study of material properties.
Keywords: impact indentation, neural network, task of classification, artificial intelligence, dynamic indentation, non-destructive testing.
The article is devoted to the use of artificial intelligence tools to solve technical problems in the construction industry. It is noted that the use of neural networks will allow taking into account the behavior of materials in various experimental conditions. The authors present a comparative analysis of approaches to neural network training, in particular, the structures of multilayer and LSTM networks are considered. It is established that LSTM networks are more effective in solving problems of identification properties of materials.
Keywords: neural network, non-destructive testing, identification task, multilayer network, LSTM network, impact indentation, indentation, strength properties of materials, neural network technologies, statistical distribution