Study on Predictive Maintenance for Controller Failure

Abstract

It is more important to be able to predict a failure before it occurs than to fix it after a failure. Attempts to predictive maintenance beyond reactive maintenance and preventive maintenance are increasing. This paper contains basic research to design a system that can predict controller failure before it occurs. The database related to the controller should be designed considering not only the data of the controller itself, but also the information on the operating environment, and the processing contents in the failure situation. In the latter part of the paper, data classification and analysis algorithms that can be used for failure prediction are summarized.