Failure Detection And Isolation Experiments With The Langley Mini-Mast

One of the technologies essential for control of large space structures is fault tolerance. A feature of control systems designed to tolerate failures of some of their components is a monitor which detects the event of a component failure and isolates the failed device. This paper reports on experiments to demonstrate failure detection and isolation (FDI) using the flexible truss facility Mini-Mast at the NASA Langley Research Center. A number of techniques for performing the FDI function have been reported in the literature. Of these, two are selected for study because they are applicable both to sensor and actuator failures, and they do not depend on hypotheses about the forms of posible failures. These two are the method of generalised parity relations and the failure detection filter. These methods utilize the concept of analytical redundancy and therefore their performance depends on the fidelity of the model of the dynamics of the system being monitored. In this paper, results are given for sensor FDI using generalized parity relations and input-output data collected during operation of the Mini-Mast. Component failures are simulated in the data. The dependence of their performance on choices of the parameters in their implementation is explored.