Modeling and compensation of asymmetric piezo-actuated flexure-based system hysteresis

• Main Author: Zhou, Chao
• Other Authors: Ang Wei Tech
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2022
• Subjects: Engineering::Mechanical engineering::Robots; Engineering::Mechanical engineering::Control engineering
• Online Access: https://hdl.handle.net/10356/157741
• Institution: Nanyang Technological University

Micro-manipulation is extremely challenging as it requires high repeatability of the system, as well as superior accuracy of the control algorithm. Piezo-actuated flexure-based systems are very popular in micro-manipulation applications, such as handheld surgical robots, auto-focusing optical systems, high precision fabrication tools and other applications. One of the biggest challenges in using these systems in dynamic applications is the undesired hysteresis. Various methods are proposed to model and compensate for the hysteresis behavior, and it is yet not well solved. Limitations of state-of-the-art literature exist in modeling and compensating the asymmetric and rate-dependent hysteresis behavior of piezo-actuated flexure-based systems. Among the popular operator-based models, the Prandtl-Ishlinskii (PI) hysteresis model stands out because it is easier to be implemented and has less computational complexity in analytical inversion. In this work, the asymmetric and rate-dependent system hysteresis behavior is studied. A dual-operators based modified PI (DPI) hysteresis model is proposed with dual-operators modeling the expansion and contraction curves respectively. Experimental results show that the DPI model reduced the RMSE by 30% compared with the modified PI model with dead-zone operators (MPI), and by 76% compared with the PI model. The most significant contribution of this work is the rate-dependent dual-operators based modified PI (RDPI) model derived from the DPI model. Using functions of rates as the properties of operators and two observers, the RDPI model can model and compensate for the asymmetric and rate-dependent system hysteresis. Experimental results show that the RDPI model can reduce the RMSE by around 24% compared with the DPI model. A controller based on the RDPI model is designed and implemented in a hand-held surgical device, enhancing the microsurgery performance by reducing the standard deviation of the stitches distribution by around 50%.