Polushko, M. M.Shevchenko, V. V.2026-06-102026-06-102026Polushko, M. M. Intelligent system for controlling surface roughness parameters of parts based on multi-task neural networks / Polushko M. M., Shevchenko V. V. // Приладобудування: стан і перспективи : збірник матеріалів XXV Міжнародної науково-технічної конференції, [Київ], 12-13 травня 2026 р. / КПІ ім. Ігоря Сікорського. – Київ, 2026. – С. 93-96. – Бібліогр. 6 назв.https://ela.kpi.ua/handle/123456789/81577In modern high-tech instrument-making, ensuring a given surface roughness is a critically important stage that directly affects the operational reliability, wear resistance and fatigue strength of parts. The traditional approach to controlling roughness parameters is usually based on post-processing analysis, when measurements are taken after the completion of the machining operation using contact profilometers or optical systems. This approach creates significant time delays and does not allow for a prompt response to deviations in the technological process, which often leads to the appearance of defects. The solution to this problem is the implementation of intelligent prediction methods that allow assessing the surface quality directly during cutting, using indirect physical features, such as vibration signals.ensurface roughness parametersneural networktool wear controlvibration signal analysisArticleС. 93-96621.9.08:658.562