Charles Explorer logo
🇬🇧

Automated system for remote defect inspection

Publication at Faculty of Physical Education and Sport |
2018

Abstract

The paper deals with development and application of a method which allows to perform automatic check the surface integrity of the monitored component. In our case, it is the bearing race of the rotating cylindrical container of the lime regeneration line in the paper mill factory.

By visual checks, some cracks have been discovered on the circular surface of the race. These cracks have needed to be monitored to prevent fatal failure.

To measure the cracks manually, it was always necessary to stop the cylinder. However, these breaks have caused momentous problems in the operation of the line, which is conceived for continuous work due to the nature of the production process.

This was the main reason for the development and deployment of the automated system. Because the problem takes of low-frequency bearing run (less than 1 Hz) moreover used in outdoor conditions an image processing technology was chosen to create a tracking system to determine the immediate crack length at specified times.

Due to the very limited time span for the development of a separate robust apparatus and with regards to the fact that the system is used only till replacing the concerned race with a new one, i.e. for a period of about three months, the hardware of the apparatus was based on a standard web camera with fullHD resolution connected to a notebook with a controlling and recording software created in the Matlab environment. This "recording" system located close to the monitored bearing is wirelessly connected to the "controlling" computer placed in the control center of the paper mill.

The system reduces the data stream of optical scanning of 750 mm wide bearing ring by cut of the recorded figures to a data level that reliably enables their further processing and evaluation. It is done by the control center computer.

The system is able to indicate individual cracks present at the bearing ring surface with image data processing. Documentation and assessment of development of the dimensions of identified and observed cracks are also presented.

For a better idea about all aspects of the problem, we add that the ring has the diameter of 4.7 m and scanning width of almost 800 mm due to additional axial motion.