Деякі аспекти класифікації технологічних методів створення робочих поверхонь деталей машин

Abstract

The paper presents a fundamentally novel approach to the classification of technological methods for forming functional working surfaces of machine parts, based on the integration of neural network models with systems analysis. The relevance of the study is stipulated by the increasing requirements to the operational properties of the surface layer—wear resistance, corrosion resistance, contact stiffness, and so on—which necessitate the development of highly adaptive intelligent systems, enabling to select automatically optimal technological methods, taking into account diverse morphological, physical‑mechanical and structural factors. The aim of this work was to develop a multilevel methodology for classifying technological processes of surface layer formation using various types of neural networks and the concept of a multidimensional parametric space. Within this methodology, each technological method is formalized as a parameter vector encompassing the process’s energy characteristics, its influence on surface microgeometry, physical‑mechanical properties, changes in structural‑phase state and techno‑economic indicators. This approach provides a clear, structured representation of the interrelationships between the characteristics of technological methods and their operational outcomes. For validation, representative parts—shafts, gears, bearing rings, and housing components — were selected, and a comparative analysis was conducted of the system recommendations, expert assessments, and simple criterion‑based models. The results showed that the accuracy of method selection in the developed system increased by 27…34 % compared to traditional approaches; wear intensity decreased by 18…42 %; and corrosion resistance of the surfaces improved by 15…35 %. Moreover, the classification speed did not exceed 1.2 s, which proved to be significantly shorter than the execution time of analogous systems based on expert rules and statistical methods. One of the key advantages is the integration with CAD/CAM/CAE environments, which enables the prediction of operational properties during the design stage. The high adaptability of the system allows for prompt accommodation of new technological solutions and changing production conditions. Thus, the proposed intelligent classification system for technological methods of forming functional surfaces represents an effective tool for automating the selection of optimal technological solutions, contributes to improving the quality and reliability of mechanical engineering products, and opens prospects for further scientific research and practical industrial implementation.