KIMM researchers implemented machine vision technology technology to conduct diagnoses of machine structure. (image: Korea Institute of Machinery & Materials)
DAEJEON, Jul. 3 (Korea Bizwire) — The Korea Institute of Machinery & Materials (KIMM) announced on Tuesday that it has developed a new machine vision technology that automatically detects failures in various machines.
Machine vision technology is a system created to imitate how humans look at things and make appropriate decisions.
The technology has been used to check photos of thousands of electronic circuit boards for any abnormal connections and mislabeled parts.
KIMM researchers implemented this technology to conduct diagnoses of machine structure.
A Convolution Neural Network, a deep learning algorithm most commonly used in the interpretation of visual images, has been used to watch various video clips and learn how a machine works, based on which it can detect vibrations and other abnormalities that occur inside a machine.
In a test run, the new machine vision showed 100 percent accuracy in distinguishing normal and abnormal vibrations of a water pump shown through a video.
The KIMM explained that the new technology will be able to replace the current methods of using multiple vibration sensors or hiring experts, both of which are either too complex or difficult to proceed.
“The new machine vision technology can be used in plant industries,” said Sun Kyoung-ho, a head researcher at the KIMM. “It can be safely used even in dangerous industrial sites where human access is difficult.”
H. M. Kang (hmkang@koreabizwire.com)
