Fault Detection and Diagnosis (FDD) is the process of automatically detecting and identifying faults in systems. A fault is an defect or abnormal condition which can cause or lead to system failure if not corrected. FDD systems offer safety benefits as well as economic benefits through preventative maintenance, task automation and downtime reduction. Research into FDD seeks to identify faults early on and detect increasingly minor fault conditions to give operators more time to address faults before failure and insure equipment operates in optimal conditions.
To tackle these challenges, researchers have been making use of more complex machine learning techniques. However, this means troubleshooting and adjusting FDD processes has also become more complex. This tutorial is intended to introduce how data separability measures can be applied to the type of problems seen in FDD research and be used to gain information about a dataset which can then be applied to troubleshooting, evaluation, testing, etc.
© Copyright. McMaster University. May be used for educational purposes without written permission but with a citation to this source.