飛機氣動力參數判別

Abstract

本論文主要是以子空間判別法(Subspace Identification)對AIAA文 獻 中所提供的簡化的飛機縱向飛行線性動態系統在作爬升動作時,對飛 機的 時變參數加以判別,使用的子空間判別有:確定性判別( Deterministic Identification),遺忘因子法(Forgetting Factor method)及狀態與輸入 的遞推算法(Recursive method),另亦使用傳統判 別中最基本最典型的遞 推最小平方法(Recursive Least Squares method)及限定記憶最小平方法 (Fixed Memory Least Squares method) 加以判別比較.期能達成飛行系統 參數的即時判別,提供即時控制使用. 本論文所發展出的遺忘因子法及狀態與輸入的遞推算法,使子空間判別 法能適用於時變參數.另外在我們模擬過程中亦發現,最小平方法與子空間 判別法單獨使用一種方法判別時在部份情況下會有不能有效判別的狀況, 因此兩種判別法都使用時可以讓不能有效判別的區域降至最低. In this thesis we mainly use the subspace method to identification the parameters of a simplfied and linearized AIAA dynamic aircraft model. We simulate the situation of climbing and identify the state-varying aircraft system in longitudinal direction. We use deterministic identification, forgetting factor method and state-input recursive method for subspace identification,and recursive least square estimate, fixed memory least squares estimate to identify this system. We expect to achieve the objective for real-time control. The forgetting factor method and state-input recursive method developed in this thesis are applicable to state-varying systems, Besides, we found that if using least squares or subspace method alone, we may encounter the unidentifiable results for some input-output data. Hence, if we use both two methods, we can increase the identifiable regions.