| 摘要: | 本研究提出「天鉤主動隔震系統」，應用於單自由度隔震機構並詳列其運動方程式及控制律，進行數值模擬分析及實驗之驗證。改良傳統天鉤控制理論，將以絕對速度回饋之控制力，調整為量測相對地表速度及地表加速度，藉由積分地表加速度以獲取地表速度訊號，再以相對速度訊號及地表速度訊號計算控制力；不僅能提升訊號量測的便利性，更能增加回饋訊號的穩定性。為了掌握地表加速度積分濾波運算成地表速度，自行引入積分濾波器，並藉由濾波消除積分導致之穩態誤差。將天鉤主動隔震系統方程式擴展為包含積分濾波器之形式，使控制及最佳化設計皆能考量積分濾波器之影響，如：時間延遲誤差、穩定性問題。由於天鉤主動隔震系統非全狀態回饋，利用直接輸出回饋最佳化(direct output feedback optimization)最佳化設計方法，以絕對加速度最小化為設計目標，設計控制力增益參數。有別於傳統天鉤控制理論，天鉤主動隔震系統藉由最佳化設計及兩個控制力控制參數，使固有阻尼能被充分考量；因此，不同固有阻尼比之系統能有一致的最佳化反應。於頻率反應函數分析中，對於一般地震的顯著頻率範圍0.1Hz至10Hz，天鉤主動隔震系統不僅優於被動隔震系統還優於傳統天鉤控制理論。於地震歷時分析中，天鉤主動隔震系統對於近域地震及遠域地震皆有良好的隔震效果，且優於被動隔震系統。於敏感度分析中，相對位移、絕對加速度、控制力均對於控制力增益參數變化敏感；另外，控制力亦對於勁度的變化敏感。於穩定性分析中，兩個控制力增益參數於一定範圍內的變化，系統恆為穩定可控制，而當中作為地表速度回饋之控制力增益參數不影響系統之穩定。藉由線性伺服滑台實現天鉤主動隔震系統，並以振動台實驗驗證其可行性。實驗結果表明，天鉤主動隔震系統對於不同地震皆有一定的隔震效果。實驗結果與數值模擬之差距，經由功率頻譜密度分析，發現是因為設備產生之高頻振動及伺服滑台對於高頻的控制有其限制所致。;In this study, the skyhook active isolation control theory is developed and applied to a single-degree-of-freedom active isolation system. The equation of motion and proposed control algorithms are derived in detail. The numerical simulation analysis and experimental verification of the developed skyhook active isolation system are carried out. The proposed control algorithm improved the traditional skyhook control by adjusting the measurements from only the absolute velocity of the system to the velocity relative to the ground as feedback signal; meanwhile, it uses the ground velocity as the feedthrough signal. The velocity relative to the ground and ground velocity can have individual control gains. In addition, the ground velocity is obtained by integration of the measuring ground acceleration. This modification not only improves the convenience of measurements, but also enhances the stability of the feedback signal. To access the ground velocity from the ground acceleration, the real-time integral-filter is introduced and designed to process the integration and eliminate the caused steady-state error. Moreover, this integral-filter can be directly combined into the model of the skyhook active isolation system, so that the influence of the integral-filter, i.e. the time delay issue or stability problem, can be guaranteed. Since the proposed skyhook active isolation system is not using full-state feedback, the direct output feedback optimization can be adapted to determine the optimal control gains with the goal of minimizing the absolute acceleration of the system. Different with the traditional skyhook control, the proposed skyhook active isolation system uses the relative velocity and the ground velocity as measurements, so that the influence of inherent damping can be fully considered. Therefore, even though the system has different inherent damping ratios, by adapting different control gains, the skyhook active isolation system can achieve the same optimum isolated responses. In the frequency response function analysis, the skyhook active isolation system is more outperforming than the passive isolation system and traditional skyhook control, especially when the frequency of base excitation is between 0.1Hz to 10Hz which is the range of dominant frequency of normal earthquakes. Furthermore, in the time-history analysis, the performance of the skyhook active isolation system is much better than the passive isolation system whether subjected to far-field earthquake or near-fault earthquake. In the sensitivity analysis, the relative displacement, the absolute acceleration, and the control force of the skyhook active isolation system are all sensitive to the control gains. The stability analysis also showed that the skyhook active isolation system is always stable, even the control gains various within a certain range. Besides, the control gain, which is the ground velocity signal feedthrough, does not affect the stability of the system. Finally, the skyhook active isolation system is practiced by a linear servo slider to verify its feasibility by the shaking table experiment. The experimental results show that the skyhook active isolation system performs well to isolate seismic force. However, there is still a certain difference between experimental results and numerical simulation. The difference is majorly existing at high-frequency range which is observed by power spectral density of the absolute acceleration. Therefore, this high-frequency vibration is considered caused by the friction of the ball screw mechanism rather than the control algorithm. |
