An Experimental Study on Active Vibration Control of Flexible Beam Using Model Predictive Control

• Main Authors: Yi-Huan Huang, 黃一桓
• Other Authors: Chi-Ying Lin
• Format: Others
• Language: zh-TW
• Published: 2012
• Online Access: http://ndltd.ncl.edu.tw/handle/23874848233584691815

碩士 === 國立臺灣科技大學 === 機械工程系 === 100 === As a common example in various kinds of structural systems, flexible beam has the light damping property which easily affects the system performance and structure safety. To compensate for this drawback active vibration control is one proven-effective technique which embeds actuators and sensors into the structures to reduce the influences of disturbances in real-time. Because applying classical active vibration control methods is unable to satisfy the increasingly stringent requirements and constraints, recently researchers have started to use advanced control methods for vibration suppression of flexible structures. This research focuses on the experimental investigation on vibration control of a flexible beam using Model Predictive Control (MPC), an advanced optimal control method which can handle constraints at each sampling step. Using the Hildreth’s optimization solver, the work considers three different MPC strategies, including a basic MPC, a terminal-state involved MPC, and a MPC combined with repetitive control (RMPC), for control design and performance evaluation. The study specifically investigates the periodic disturbance rejection performance of applied MPC methods, showing the effectiveness of RMPC. To reduce the computational burden and improve the practicability of MPC in active vibration control applications, this study also applies a recurrent neural network as fast optimization solver, to implement the aforementioned MPC strategies. Besides the parameter analysis, the experimental results particularly show the constraint handling and performance improvement by considering the input constraints in the MPC design. Finally, this thesis summarizes several concluding remarks on control parameters selection as a guideline for future designers.