Analytical Model-based Fault Detection in Air-launched Systems Using Statistical Markov Models

In this paper, a model-based fault detection problem for air-launched systems is considered. Firstly, the position, velocity and attitudes of the system are estimated using Kalman filter, then the sensor fault is detected by defining a suitable threshold. The fault detection is done using stochastic forward variables. In this algorithm, the covariance of data is used to model the faulty mode of sensor. The Monte-Carlo simulations was used to adjust the parameters of the algorithm in static mode. Numerical experiments on an unmanned aerial vehicle show when the system states are observable the fault detection algorithm is capable to detect the sensor fault. When the system lost its observability condition, the algorithm just can detects the instantaneous faults but, the method cannot detects in the static mode.