The researchers say that the embedded Internet of things sensors can not only monitor mechanical failures, but also allow machines to repair them automatically.
Due to the application of the industrial Internet of things (IIoT), a lot of downtime has been reduced. Scientists believe that embedded sensors and a large amount of data collected can create "self - Healing" manufacturing equipment.
A passionate SelSus project is based on this concept. At present, many European academic institutions and manufacturers (including Ford Motor Co) have been involved in the exploratory research of this project.
The idea of team is not only to find faults in production process, but also to automatically repair potential problems through self repair program based on mathematical algorithm. The scientists say that equipment diagnosis should be carried out before the equipment failure, and the "self healing" ability will make the equipment monitoring to a new height.
Study on the German Fraunhofer manufacturing engineering and automation (which is the SelSus project in one of the institutions) Martin Kasperczyk said in a news release: "this project is not only to monitor the running status of machines and components, can also make the system capable of early detection of potential weak points or signs of wear, so that the system can predict early potential failure."
Even in some cases it should be able to automatically repair these failures.
The team has made some progress. One of the systems developed by the European Union funded research partners is used for the production of the engine arm, which can repair itself in case of failure. "If it monitors resistance, it will automatically avoid it without breaking itself," the researchers said.
The system can also predict the possibility of power failure of the cable under load.
How "self healing" technology works
The network sensor drive technology is mainly based on the Bayesian network mathematical model of computing probability. It works with learning software to analyze how the machine works and is part of the combination of algorithms.
It's not easy.
"There are several algorithms that can't be achieved by programming," Kasperczyk said.
Model data are acquired on machine installation and tested under load, and then compared with actual operation conditions, and alarm is provided to man-machine interface.
Fraunhofer IPA's Michael Kempf said on the project website: "the main challenge is to establish decision models and simulation models to reflect the real manufacturing environment."