2019 IEEE 15th International Conference on Automation Science and Engineering (CASE) | 2019
Detecting Intermittent Faults with Moving Average Techniques
Abstract
So far, the problem of detecting intermittent faults has not been fully investigated in the multivariate statistical framework. The intermittent fault is a kind of non-permanent fault which lasts within a limited period of time and then disappears itself. Generally speaking, intermittent faults have small magnitudes and short durations, which consequently makes them even more difficult to detect than incipient faults and causes losing efficacy for many existing methods. In this paper, a generic quadratic-form index, combined with the moving average technique, is utilized to detect intermittent faults. In order to determine the window length, fault detectability is studied and a sufficient condition is derived. Moreover, compared with the permanent fault case, false alarms have a worse influence on the detection of intermittent faults. Thus, a method to exclude false alarms is proposed based on the detectability analyses. Finally, a numerical simulation is carried out to verify the theoretical results and illustrate the effectiveness of the method.