Sensor Failure Detection within the TBM Framework: A Markov Chain Approach

V. Ricquebourg, D. Menga, M. Delafosse, B. Marhic, L. Delahoche, A.M. Jolly-Desodt.

This paper presents a sensor failure detection method based on the fusion of predicted and observed sensor data. The originality of our approach is the use of a Markov Chain to model the normal behavior of a sensor within the TBM framework. When fusion between predicted and observed data is done, three experts analyze conflict resulting from the fusion process and are able to detect an abnormal behavior of the sensor by looking for high increase of conflict. The testing results show that this method is efficient to detect sensor failure with a TBM approach.

PDF full paper