Detecting and identifying multiple failures in a flight control system

Abstract

Parity space and parameter estimation methods are combined to detect and identify sensor and actuator failures in a control system. Parity residuals (or error indicators) are generated from analytical relations describing the dynamics of the system. Representing the failures in sensors and actuators as biases and changes in scale, a recursive least-squares estimator is designed to identify the source and magnitude of all failures from the information contained in the residuals. The choice of sensors, length of the information window, and possible combinations of sensors and actuators on the performance of the failure detection and identification procedure are addressed. Simulation results on a linear aircraft model with wind disturbances and sensor noise show the validity of the present approach.