英文摘要 |
Seismic shake table testing is one of the most straightforward methods for performance assessment of structural resistance under earthquakes. Since the total weight of shake table and testing specimen is enormous and close to the maximum force capacity of the hydraulic actuator, the control performance of shake table testing is greatly affected by the interaction between the shake table and actuator, leading to difficulties of accurate acceleration control of shake table. Various control strategies such as Proportional-Integral-Derivative, and Three-Variable Control are applied to the seismic shake table testing to overcome the situation. Recently, there were breakthrough in image recognition, audio processing, image generation, and robotic control with deep learning approaches. In this study, deep learning is adopted to improve the control performance of shake tables. The strategy and future work are introduced thoroughly. |