Multi-Mode Control of Structure Using Neural Networks

應用類神經網路於結構之多重模態控制

楊世銘(S. M. Yang) ； 林俊男(J. N. Lin)

《中華民國振動與噪音工程學會論文集》 1996 (1996/04) Pp. 215-227

https://doi.org/10.30028/CSSVANNUAL.199604.0215

摘要

智慧結構乃指結構物內含有感測器與致動器，可以隨環境需要而主動地改變結構外形與物理性質。此研究旨在應用類神經網路於智慧結構的系統識別與多重模態振動控制。本文提出二個類神經網路，其一由系統識別以建立結構系統之動態模式，另一則由振動抑制的性能要求以線性參考系統來建立控制器。實驗結果顯示此類神經網路均能有效地識別系統之前兩個自然頻率11和53Hz。同時利用預埋於智慧結構內的壓電致動器可成功地抑制前兩個振動模態。結構系統受週期外力或隨機外力之擾動時之振幅均可以被有效地消除達60%與50%。

並列摘要

The system dynamics of composite smart structures are often difficult to analyze because the embedded piezoelectric sensors and actuators, system identification become indispensable for effective vibration control applications. This paper presents an integrated approach of identification and vibration control of smart structures by using neural networks. Two neural networks are developed, one identification for system modeling and the another controller network for the vibration suppression. It is shown experimentally that the neural networks can faithfully identify the first two vibration modes at 11 and 53 Hz. The neural controller can also suppress the vibration of these two modes simultaneously by using the piezoelectric actuator embedded in composite smart structure. The closed-loop responses under two-mode harmonic excitation and under random excitation are reduced by 60% and 50%, respectively.

並列關鍵字

Neural Networks ； Smart Structures ； Vibration Control