結構物裝置非線性阻尼器之系統識別研究

Application of Feature Extraction and Identification Technique to Building Structures Implemented with Passive Dampers

劉建榮(Chien-Jung Liu)

指導教授 ： 羅俊雄

國立臺灣大學/工學院/土木工程學研究所/碩士(2013年)

https://doi.org/10.6342/NTU.2013.02573

摘要

本文研究的目的在於討論一個簡單之結構物裝置非線性阻尼器如磁流變阻尼器( Magnetorheological damper，MR damper )後了解此阻尼器對於一簡單結構之系統動態特性之改變量，並且了解阻尼器提供多少勁度、阻尼，並且裝置的位置也是考慮因素之一，對於裝在一樓及一、三樓及一、三、五樓是否裝置阻尼器越多，非線性之現象越明顯。首先從實驗量測之反應歷時、各個樓層之遲滯迴圈、各樓層之最大層間位移反應資料進行初步分析。利用直接量測之結果對此結構有初步了解後。接著使用非參數法如主成分分析法( Principal Component Analysis)計算出系統之各個模態與各個模態之模態貢獻量。用小波分析法( Wavelet Packet Transform )繪製出頻率隨時間之變化如希爾伯特邊緣頻譜(Hilbert Marginal Spectrum)，與組成訊號之組成能量( Wavelet Packet Node Energy )了解頻率隨著加入越多的阻尼器對於結構物之影響。還有假設系統受到非線性之行為作用不大，因此把系統假設成等值線性。利用如子空間識別法(Subspace Identification, SI)，繪製出穩態圖找出系統之動態特性後，探討系統之動態特性有哪些變化。使用自我回歸外變數之模型識別法( ARX-Model )繪製出各個樓層對地表之轉換函數，利用轉換函數( Transfer Function )之變化來了解阻尼器之影響。最後計算出每個樓層阻尼與勁度在受到地震力作用下對於此系統所消散之能量。此外把子空間識別法應用在實際收集資料如新土木研究大樓受到地震力作用下，探討中間隔震層對於土木研究大樓之動態特性之改變。

關鍵字

子空間識別法 ； 自我回歸外變數之模型法 ； 磁流變阻尼器 ； 希爾伯特邊緣頻譜

並列摘要

In this research, many algorithms were applied to identify the dynamic characteristics of building structure implemented with passive control devices (using MR-dampers with constant voltage) during earthquake excitation. And also want to understand the passive control devices provide how much effect to the building. Also check the effect of the location of the control devices.
First use the measured data like absolute acceleration and story drift to understand the fundamental phenomenon, and draw the Hysteresis Loop to find the damper effect. Also use the algorithm like principal component analysis to find the model contribution and model mode shape. Use the Wavelet Analysis like Wavelet Packet Transform to calculate the component energy to understand the model characteristics like frequency change. Also draw the Hilbert Amplitude Spectrum and Hilbert Marginal Spectrum to see the frequency change during the time change.
Finally, under assumption of equivalent linear system, use the Subspace Identification (SI) technique to fin the model characteristics of an nonlinear system. First, through numerical simulation of a SDOF system with nonlinear damper, the results from these two approaches identification show different physical meanings. If equivalent linear system model is used, the result of the identified system natural frequency is inconsistent with the general trend of the hysteresis loop of the analysis time window. Otherwise, if the SI technique is uses, the identified stiffness is more similar to the tangent stiffness of the hysteresis loop. Application of this result to MDOF system is discussed. And use the ARX model to draw the transfer function to see the change of the transfer function. Here use the error index (EI) to find the optimize parameters in ARX model . Here also use the SI algorithm into real measured data like new Civil Engineering Research Building in the campus of National Taiwan University during an earthquake occurred.
Last, calculate the dissipation in each floor to find the damper effect to the floor when building during a earthquake.

並列關鍵字

Subspace Identification ； ARX model ； MR damper ； Hilbert Marginal Spectrum

參考文獻

[21] 羅俊雄，翁健煌(2010)，子空間識別法於系統識別及結構損壞診斷之應用，國立台灣大學土木工程學系博士論文。

[20] 羅俊雄，劉奕成(2011)，協方差型隨機子空間識別法之應用，國立台灣大學土木工程學系碩士論文。

[17] 羅俊雄，辜琪媜(2005)，利用小波轉換技術於結構振動訊號之解析，國立台灣大學土木工程學系碩士論文。

[18] 羅俊雄，陳奕安(2007)，應用小波分析於鋼筋混泥土結構物的損壞識別，國立台灣大學土木工程學系碩士論文。

[19] 羅俊雄，盧恭君(2010)，開發智慧型結構系統於土木基礎建設：感測與控制，國立台灣大學土木工程學系博士論文。