Specialists from Perm Polytechnic have created an intelligent system for defectoscopy of turbine blades, which automatically detects defects during polishing and corrects the process. This technology, based on neural network video analytics, identifies scratches, chips, and other surface damage in real time with an accuracy of up to 96%. This eliminates errors associated with the human factor and significantly speeds up the production process.
Turbine blades are a critical component for aviation and energy systems. Their quality directly affects the reliability and efficiency of engines. Currently, the condition of these parts is monitored manually. The operator carefully inspects the surface of the blade after polishing, which requires significant time and is associated with the risk of missing minor defects.
Our complex allows not only to detect the smallest flaws, but also to automatically optimize the processing process. This significantly increases the accuracy and speed of production.
The prototype of the system has already been tested at a motor-building enterprise and showed a recognition accuracy of 96% and a completeness of 94%. In the future, the developers plan to expand the database, introduce 3D modeling of blades, and scale the system for use in various industries.
The introduction of technology into industry will improve the quality and competitiveness of products, reduce control time, and reduce the cost of rejects and rework of products.
Read more on the topic:
Russia plans to develop a new aircraft model using the An-124 as a basis
Historical flight: Sukhoi Superjet 100 will connect Khabarovsk with the Philippines
Ural Alaska: How Putin's gift inspired the Irbit plant to create exclusive motorcycles