Development of Data-Driven Model and Control Techniques for a Two-Link Flexible Manipulator (TLFM)

Abstract

There has been an ongoing interest in the modeling and control of flexible manipulators (FMs) due to their potential application in many domains such as space, biomedical and industrial fields. But because of the vibration effect of FM links, different modeling methods and control techniques are studied to improve the system performance. A two-link flexible manipulator (TLFM) has complex math models such as the assumed mode method (AMM) model and a simplified lumped parameter method (LPM) model, which are used to design different control techniques. This paper aims to develop a linear data-driven model (LDDM-1) of a TLFM, including a comparative study of this model with other models. LDDM-1 is developed using the input-output data acquired from Quanser 2-DOF serial flexible manipulator (2DSFLM) but with a model structure based on the LPM model. The other TLFM models, LPM and linear data-driven model 2 (LDDM-2) are compared with LDDM-1 while using the response of the physical system as the benchmark. Besides, PID control, fuzzy logic control (FLC) and LQR control are designed. Then, the closed-loop simulations of the controllers with different models are conducted in MATLAB/Simulink. After simulation, the controllers are implemented on Quanser 2DSFLM experimental platform. The simulation and experimental results indicate the performance of FLC is better than LQR and PID based on the position control and tracking response of the manipulator. At the same time, LQR outperformed the other controllers in deflection suppression.