The problem considered in this dissertation is the design of a failure detection, estimation, distinction, and compensation (DEDC) system. This failure DEDC system provides adequate protection against failures which could occur at unknown time, unknown location, unknown type, and unknown magnitude.; A Kalman filter is employed to generate residuals which are referred to observations. A hypothesis test is used to detect whether or not a failure exist. Recursive estimation algorithms are derived to estimate the size of failure. The parallel distinctor is used to distinguish between the type of failures. Then, by reconfiguring the system, the failure in the system is compensated.; The procedure is applied to two examples, and the results are presented and discussed.
展开▼
