本論文針對具有結構不確定性 (structured uncertainty) 之線性系統，
考慮其可能發生之感測器或制動器之故障以及系統不確定性之影響，發展
具備強健性之故障偵測與鑑別之方法。所產生之強健殘餘函數具有偵測系
統故障之能力，但不會受系統不確定性之影響，依據該函數可偵測故障發
生之時刻並鑑別故障之程度。此方法利用部份狀態估測器 (partial
state observer) 以比文獻直截了當之方式推導殘餘函數，並改善文獻中
以未知輸入估測器 (unknown input observer) 為主之故障偵測方法之缺
點。論文中並闡述此函數存在之充分必要條件以及其設計步驟，並以垂直
起降之直升機作為模擬實例，俾驗證其有效性。由模擬之結果可知，此方
法對於突發故障或和緩發生故障之感測器與制動器，兼具有同時偵測之能
力。

This study aims to develop a robust method to deal with fault
detection and identification (FDI) for linear system subject to
structured plant uncertainties or parameter variations. The
approach consists of the use of a partial state observer and
an augmented system accounting for possible sensor or actuator
failures as well as plant uncertainties. The generated robust
residual signals are sensitive to faults to be detected while
insensitive to the plant uncertainties. It therefore serves as
a fault indicator. The sufficient and necessary condition for
the existence of robust FDI and its design procedure are
presented; and the application is illustrated with the help of
simulation studies performed on a vertical takeoff and landing
helicopter. Superior to previous methods in the literature,
the proposed method is able to classify the sensor and
actuator failures simultaneously in the presence of both abrupt
and smooth bias faults.