Rostec's United Engine Corporation has launched an industrial complex for inspecting aircraft engine blades based on artificial intelligence at the ODK-Saturn enterprise. The "Control Point" system combines robotic part feeding, ultra-high resolution cameras, and neural network models. A robot independently feeds the blade into the inspection area, and a camera scans the surface – AI detects deviations from the norm as small as 40 microns and non-contact applies marking indications.
Before the system's implementation, each blade was manually inspected by a specialist from the technical control department. This created a risk of missing defects due to human factors and became a bottleneck as production volumes increased. Now, two automated stations are in industrial operation and provide preliminary control for six types of gas turbine engine blades.
Evgeny Alekseev, Digital Transformation Director at ODK-Saturn, noted that AI took on the most labor-intensive part of preliminary control, and the system's accuracy metrics have reached target values and continue to grow.
Plans include scaling the solution to other types of parts and ODK enterprises. Alexander Ganin, CEO of "Tochka Zreniya" company, emphasized that bringing a solution for controlling microscopic indications on curved mirror surfaces to industrial operation was a significant milestone.
In fact, this is a transition from manual control to predictive analytics: data on each defect is saved for analysis, which allows identifying the causes of recurring inconsistencies and eliminating them before defects appear.
Earlier, www1.ru reported that neural networks were entrusted with checking PD-8 and PD-14 aircraft engine blades.