利用弱型式之數值微分運算子重建非線性系統之外力

To recover external forces of nonlinear systems using weak-form numerical differentiators

林政璋(Jheng-Jhang Lin)

指導教授 ： 劉進賢

國立臺灣大學/工學院/土木工程學研究所/碩士(2016年)

https://doi.org/10.6342/NTU201601896

摘要

二階微分之噪音訊號被寫為二階常微分方程式，當作一種特殊實例未知外力在二階線性系統之回復，轉換成線性拋物線型的偏微分方程式。之後運用格林第二恆等式推導出以伴隨崔維茲頻譜函數表示之邊界積分方程。我們發現一種弱型式的方法可回復外力，之後，從給定之噪音中發展弱型式二階運算子(WFSOD)計算訊號的二階導數，只有訊號本身是具體指定，不需要此噪音訊號之一階導數，最後，在大時間區間和大噪音之下使用弱型式方法回復非線性系統之外力。

關鍵字

弱型式二階微分運算子 ； 非線性反算振動問題 ； 伴隨崔維茲測試函數 ； 格林第二恆等式 ； 弱型式方法 ； 頻譜函數

並列摘要

The second-order differential of a noisy signal is written as a second-order ordinary differential equation, being a special case of the recovery of unknown external force in a second-order linear system, which is transformed into a linear parabolic type partial differential equation. Then the Green second identity is employed to derive a boundary integral equation in terms of the adjoint Trefftz spectral functions. We find a weak-form method to recover the external force and then a weak-form second-order differentiator(WFSOD) is developed to compute the second-order differential from a given noisy signal, of which only the signal itself is specified, without needing of its first-order differential. Finally, the weak-form method is used to recover the external forces of nonlinear systems within a large time interval and under a large noise.

並列關鍵字

weak-form second-order differentiator ； Nonlinear inverse vibration problem ； Adjoint Trefftz test functions ； Green’s second identity ； weak-form method ； spectral functions

參考文獻

[27] 陳柏穎：利用李群微分代數方程法即時重建作用於非線性結構之外力，國立台灣大學碩士論文。(2014)

[45] 蔡孟芹：化為偏微分方程的李群微分代數方法識別非線性結構外力，國立台灣大學碩士論文。(2015)

[6] CS Liu,Chang JR,Chang KH,Chen YW. Simultaneously Estimating the Time-Dependent Damping and Stiffness Coefficients with the Aid of Vibrational Data. CMC:Comput Master Cont 2008;7:97-107.

[41] Liu, C.-S. (2008c): A Lie-group shooting method for simultaneously estimating the time-dependent damping and stiffness coefficients. CMES: Computer Modeling in Engineering & Sciences, vol. 27, pp. 137-149.

[1] CS Liu, Atluri SN. A Fictitious Time Integration Method for the Numerical Solution of the Fredholm Integral Equation and for Numerical Differentiation of Noisy Data, and Its Relation to the Filter Theory. CMES:Comput Model Eng Sci 2009;41:243-261.