TY - JOUR
AU - Lin, Weiyang
AB - In this paper, a decoupled and adaptive polishing system without force sensors is designed for the industrial robot to track the rapid change of the contact force and eliminate dynamical nonlinearities in the polishing process. An identification method is proposed to obtain the dynamic model of this system. The system dynamic model is composed of the nominal linear model and the nonlinear model. The parameters of the system linear time-invariant (LTI) model is identified by frequency domain response with the disturbance observer in this paper. Regarding the high order differential terms and uncertain errors of the nonlinear part of the dynamic model, the Long-Short Term Memory (LSTM) is introduced for identifying system nonlinear characteristics. The bounds of the learning rate are discussed and the LSTM stability analysis result shows that the proposed method holds the Lyapunov stability. Finally, the experimental results show that a more accurate dynamic model can be established by combing frequency domain response and LSTM.
TI - Dynamic Model Identification for Adaptive Polishing System
JF - "International Journal of Control, Automation and Systems"
DO - 10.1007/s12555-021-0205-y
DA - 2022-09-01
UR - https://www.deepdyve.com/lp/springer-journals/dynamic-model-identification-for-adaptive-polishing-system-3LibC8qk09
SP - 3110
EP - 3120
VL - 20
IS - 9
DP - DeepDyve
ER -