Integrity Analysis and Fault Detection of Flight Test Data

Abstract

Undetected sensor malfunctions during flight testing can lead to cost overruns and program delays. Determining the presence of these faults in a timely manner allows the operator to mitigate their effect. One way to detect these faults is to use a priori knowledge of sensor calibration data and system dynamics to calculate measurement uncertainties. These can then be used to determine the integrity of the sensor data and report violations of expected sensor behavior. Analytical redundancy methods and residual processing can be used in conjunction with a priori sensor information to detect faults otherwise unobserved with single-instrument data as well as to isolate and identify failure modes. These simulation and analysis methods have been implemented as MATLAB® Simulink® blocks and were used to model the flight instruments, detect anomalies in the navigation instruments, and locate the origin of the errors of a flight test data set.