訊號分解與特徵萃取技術：應用於敏感設備之減、避震

Signal Decomposition and Feature Extraction Techniques: Application to Seismic Mitigation of Vibration-sensitive Equipment

黃謝恭(Shieh-Kung Huang)

指導教授 ： 羅俊雄

國立臺灣大學/工學院/土木工程學研究所/博士(2019年)

https://doi.org/10.6342/NTU201901242

摘要

非結構元件（Nonstructural Component）的損傷將會導致功能的缺喪、財產的損失與生命安全的威脅，更有甚者可能會直接或是間接地造成無法估計的經濟損失。近年來，研究者發現僅有在非結構元件達到與結構元件相等或是更高耐震水平的情況下，整體的地震危害度（Seismic Hazard）才能有效的降低，因為過去的報告顯示，許多在震後勉力維持結構安全的建築物最終仍在地震過程中喪失了整體的功能性，並造成了損失。為了確保非結構元件在地震過程中與地震過後保持功能性的完整，以隔震系統為原型的被動裝置一直以來都被視為對非結構元件最有效的減震與耐震方案，尤其是外型以方形或塊狀為主的重要設備物，但是透過隔震系統達到減震與耐震的重要設備物仍對低頻響應較無招架之力，例如震動敏感設備物（Vibration-sensitive Equipment）若受到近斷層地震時將容易產生過高的隔震位移並導致隔震系統失效，亦或者是極端敏感設備物（Ultra-sensitive Equipment）若受到遠域地震（Far Distant Earthquake）傳遞至台灣的極微小震波時容易產生不當的相對位移並導致整個設備故障、失靈或停機。綜觀所述，為了發展對抗地震的一套完整方案，震動敏感設備物的減震與耐震是不可缺少且急需培養的技術。
為了達到此一願景，適當設計的隔震系統仍是不可或缺的一環，除外本研究將智慧的主動控制系統（Active control System）與半主動控制系統（Semi-active control System）整合至舊有之隔震系統，並且考量地震為非平穩過程（Nonstationary Process），應用先進的訊號處理技術於舊有控制方法之中，本新式控制技術將克服舊有控制方法的諸多缺點，並在地震時同時並有效的控制震動敏感設備物與結構物的反應，此外，各式控制系統與控制方法將在本文中透過數值模擬與實驗驗證兩種不同的方式探討在近斷層地震的作用下的效果，並依據不同的控制情境深入討論新式控制技術的優勢。有鑑於部分震動敏感設備物即便在設計良好的控制系統下仍無法對抗任何大、小地震，此類極端敏感設備物甚至無法在遠域地震的極微小震波下保持運作，另一種可行的方案為利用早期預警（Early Warning）的方式在震波襲擊前關掉此設備物以避免損失，為此，本研究透過調查台灣強地動觀測網（Strong Motion Network）所收集的震波與極端敏感設備物所擁有的動態特性，並與科技廠房的損傷報告進行交叉比對，最終利用特徵萃取技術有效的提取出導致停機的特徵因子，並探討利用早期預警避免停機的可能性。最後，藉由應用新式控制技術於震動敏感設備物並提供早期預警給極端敏感設備物，使得震動敏感設備物的減震與耐震得到有效的提升，重要設備物便可於地震下得到保護，進而減低整體的地震危害度。

關鍵字

訊號分解 ； 特徵萃取 ； 敏感設備物 ； 隔震

並列摘要

Nonstructural damage brings serious threats to life safety, property loss, and functional loss and results in countless economic losses, directly and indirectly. Recently, researchers have found that overall seismic hazard can only be reduced once the nonstructural components (secondary structures) receives the same degree of consideration as primary structures. To be functional during and after an earthquake, a passive control device in the form of isolation system has been acknowledged as an effective method for the nonstructural components, especially the critical equipment. However, the equipment with isolation system may still suffer from low-frequency resonances; for example, the vibration sensitive equipment may have excessive isolator displacements when subjected to a near-fault earthquake, and moreover, the ultra-sensitive equipment may have glitch and breakdown when subjected to a long-period ground motion generated by a far distant earthquake. Clearly, the seismic mitigation of vibration-sensitive equipment is one part of a comprehensive measure against earthquakes and shall be developed.
To achieve this, the isolation systems with adequate periods are still necessary. First, the smart and adaptable active or semi-active control systems are integrated with the equipment isolation systems. Considering the non-stationary nature of earthquake excitations, signal processing techniques are utilized to enhance the traditional control algorithms. This advanced control algorithm can control both structures and nonstructural components (vibration-sensitive equipment) simultaneously to overcome the limitations of traditional control algorithms under seismic excitations. Moreover, the active or semi-active isolation system combined with various control algorithms against near-fault earthquakes is investigated through both numerical simulation and experimental validation. Nevertheless, some equipment is too sensitive to survive any seismic ground motion, even the small one generated by a far distant earthquake. For this kind of ultra-sensitive equipment, an alternative measure is to turn off the equipment before seismic vibration by applying feature extraction techniques to the ground motion data collected from strong motion networks in Taiwan. The ground motion features efficiently extracted from the seismic waveforms and the dynamic characteristics numerically simulated according to the ultra-sensitive equipment are then correlated to investigate the possibility of issuing a timely warning, turning off the equipment, and avoiding the impact before the arrival of the main shock. Finally, by providing the advanced control algorithm and the informative early warning, the seismic vibration of vibration-sensitive equipment can be mitigated, the functionality of critical equipment can be secured, and the overall seismic hazard can be consequently reduced.

並列關鍵字

Signal Decomposition ； Feature Extraction ； Vibration-sensitive Equipment ； Isolation

參考文獻

Abad, M. R. A. A., Moosavian, A., and Khazaee, M. (2016). "Wavelet transform and least square support vector machine for mechanical fault detection of an alternator using vibration signal." Journal of Low Frequency Noise, Vibration and Active Control, 35(1), 52-63.

Addison, P. S. (2017). The illustrated wavelet transform handbook: introductory theory and applications in science, engineering, medicine and finance, CRC press.

Adeli, H., and Saleh, A. (1999). Control, optimization, and smart structures: high-performance bridges and buildings of the future, John Wiley & Sons.

Aldrich, C., and Barkhuizen, M. (2003). "Process system identification strategies based on the use of singular spectrum analysis." Minerals engineering, 16(9), 815-826.

Alhan, C., and Gavin, H. P. (2005). "Reliability of base isolation for the protection of critical equipment from earthquake hazards." Engineering Structures, 27(9), 1435-1449.