Abstract
The problem of detecting sensor faults in the presence of modelling errors is addressed. In order to avoid false alarms, a statistical test is proposed in which the effects of undermodelling are properly taken into account using a stochastic embedding technique. The test concentrates on checking whether a weighted sum of prediction errors exceeds a particular threshold. The effectiveness of the test is demonstrated on a laboratory test rig.
Full Text
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call