Zagorsk Pipe Plant has implemented a pipe quality control system with computer vision algorithms

2021: Commissioning of Pipe Quality Control System

On May 24, 2021, Zagorsk Pipe Plant announced the launch of a system using computer vision and deep machine learning methods, which significantly simplifies and optimizes the control of production technology. Software and hardware complex is deployed in the area of application of external coating. The partner in the development of the solution and its integration into the ZTZ production system was the Point of View company (Yaroslavl).

Zagorsk Pipe Plant Introduces Artificial Intelligence Technologies for Product Quality Control

As part of the implemented project, more than ten chambers were installed in the coating workshop, video surveillances which fix the production process in open sections of the line. Video data from cameras comes to servers enterprises where they are processed algorithms computer using vision, which separate individual units of production from the video stream.

The following intelligent algorithm allows you to "accompany" found objects, that is, determine on which section of the line each particular pipe is at a given time. An important feature of the system is the calculation of the pipe position in closed areas. Thus, it is possible to trace each individual pipe in a common flow.

The system allows you to calculate the speed of rotation and movement of pipes, as well as link the values ​ ​ of process parameters - the readings of sensors installed on the line - for example, heating sensors, insulation thickness and others with which this pipe was produced.

All information goes to the production IT system and is analyzed by quality specialists. If any deviations are detected, the corresponding pipes are removed from production and sent for revision.

"The company produces large-diameter pipes for large customers, including state corporations. This is a responsible product with very high quality requirements. We pay great attention to this issue and constantly test technological solutions that help make quality control more comprehensive and deep. The proposed solution based on artificial intelligence methods allows you to determine quality indicators in relation to a specific point, to a specific unit of production and to distinguish those on which deviations from technological standards can potentially be present. Using the developed system, quality specialists can detect and analyze in detail exactly such samples, "said Evgeny Zaitsev, Director of Information Technology at the Zagorsk Pipe Plant.

A feature of the implemented solution is the development of computer vision algorithms specialized to the production conditions for ZTZ. They minimize the impact of interference - for example, moving people and equipment or changes in lighting - on the operation of the artificial intelligence system and ensure reliable tracking of the pipe throughout the production process.