Sensor/Actuator Fault Detection Under Uncertain Measurement Errors

The noncentral Wishart matrix trace-based fault detection statistic is proposed for the detection of sensor/actuator faults under uncertain measurement errors in this study. This approach, in contrast to the majority of current innovation-based fault detection techniques, enables the detection of sensor/actuator problems in the presence of additive measurement errors. The presented method uses the trace of the noncentral Wishart matrix as the fault detection statistic. To verify the fault detection statistic, a method for determining the confidence interval for the trace of a noncentral Wishart matrix is proposed. The presented approach to detect sensor/actuator faults is applied to the dynamics of an unmanned aerial vehicle (UAV). The continuous sensor bias, measurement noise increment, and actuator loss of control effectiveness type of faults are considered. The proposed and two traditional statistics for detecting faults in the presence of slowly accumulating gyroscope drifts are discussed and compared.