Blog Fostron is building a self-learning line.

The division of Single-purpose Machines at MOTORU JIKOV Fostron a.s. is developing a sophisticated control line for MOTOR JIKOV Slévárna a.s., specifically for the Pressure Foundry division. This innovative system, integrated with an artificial neural network, is designed to autonomously learn and detect defective parts. The total investment in this cutting-edge line amounts to twelve million CZK. It will focus on inspecting components destined for electric vehicles, particularly water pump covers supplied to Aisin, which will then be delivered to brands like Volvo and Jaguar Land Rover.

The primary advantages of this line are its high productivity and ergonomic design. Initially, operators will load the inserts with parts, while the system dynamically retrieves castings. Following cleaning with a blast cleaner that removes electrostatic charges, the parts undergo defect and surface quality detection using Cognex's neural network, which effectively identifies and classifies imperfections.
The process continues with flatness measurement using a 3D scanner and surface inspection. Subsequently, a pressure test and marking of the component are performed. Finally, the castings are sorted on a conveyor into four different boxes based on the type of defect.

This control line also features easy modifications to accommodate various casting types in the future, providing a significant advantage for expanding production capabilities. The water pump casting project is projected to nearly double revenues, reaching an estimated annual turnover of around two million euros with Aisin.