Reduction of Tracking Error/Contour Error of Multi-Axis Motion Stage Based on Learning Control Schemes

  • 郭 威良

Student thesis: Doctoral Thesis


Reducing tracking error and contour error is crucial for contour following applications of a multi-axis motion stage This is because contour following accuracy is most likely to be affected by factors such as external disturbances and nonlinearities In order to cope with this problem this dissertation proposes three new motion control schemes that exploit the paradigms of learning control approaches such as Iterative Learning Control (ILC) Cerebellar Model Articulation Control (CMAC) and Reinforcement Learning (RL) In the first motion control scheme proposed in this dissertation the ILC combined with CMAC is employed to suppress adverse effects due to nonlinearity and periodic external disturbance In the second motion control scheme proposed in this dissertation the cross-coupled control scheme with a parameter-based contour error estimation algorithm is employed to cope with modeling uncertainty and dynamics incompatibility among different axes Moreover to further reduce tracking error the learning rate of the fuzzy-logic-based reinforcement ILC also employed in the second motion control scheme can be self-tuned based on the tracking error information In the third motion control scheme proposed in this dissertation a reinforcement Q-learning combined with ILC is employed to adjust the contour following feedrate so as to reduce tracking error Experimental results of several contour following experiments conducted on a biaxial motion stage are used to assess the effectiveness of the proposed motion control schemes
Date of Award2017 Aug 3
Original languageEnglish
SupervisorMing-Yang Cheng (Supervisor)

Cite this