Iterative convergence control method for planar underactuated manipulator based on support vector regression model
- PDF / 861,899 Bytes
- 14 Pages / 547.087 x 737.008 pts Page_size
- 12 Downloads / 208 Views
ORIGINAL PAPER
Iterative convergence control method for planar underactuated manipulator based on support vector regression model Ya-Wu Wang · Hui-Qing Yang · Pan Zhang
Received: 10 December 2019 / Accepted: 20 November 2020 © Springer Nature B.V. 2020
Abstract An iterative convergence control method (ICCM) based on the support vector regression (SVR) is proposed to realize the position–posture control of the planar four-link underactuated manipulator with a passive second link. Firstly, the particle swarm optimization (PSO) algorithm is used to obtain the target angles of all links according to the position–posture control objective. Then, two prediction models for the coupling relationship between the first link and the passive link, and the third link and the passive link are established based on the SVR, whose optimal parameters are selected by the chaos particle swarm optimization (CPSO) algorithm. By repeatedly controlling the first link or the third link to rotate an angle which is calculated by the trained SVR model, the passive link gradually converges to its target angle after several iterations. Next, the active links are controlled to rotate to their target angles with low speeds, and the passive link does not rotate due to friction. Finally, the experimental results verify the effectiveness and feasibility of the proposed method.
Y.-W. Wang (B) · H.-Q. Yang · P. Zhang School of Automation, China University of Geosciences, Wuhan 430074, Hubei, China e-mail: [email protected] Y.-W. Wang · H.-Q. Yang · P. Zhang Hubei Key Laboratory of Advanced Control and Intelligent Automation for Complex Systems, Wuhan 430074, Hubei, China
Keywords Underactuated manipulator · Position– posture control · Support vector regression · Particle swarm optimization algorithm · Coupling relationship
1 Introduction In recent years, the control of the underactuated manipulator (UM) [1–3] has attracted extensive interest of researchers. The UM can be divided into the vertical UM which is influenced by gravity and the planar UM which is unaffected by gravity. For these two types of the UMs, the research on the control of the vertical UM [4–6] is relatively mature, while the control of the planar UM is still an open research, especially in experimental research. Many scholars have studied the control problem of the planar UM [7–9]. For the holonomic UM, Lai et al. [10] presented a control strategy to realize its position control based on its complete integrability. For the firstorder nonholonomic UM, a model reduction method is proposed in [11] to achieve the position control of the planar passive–active–active (PAA) UM. Moreover, Wang et al. [2] proposed a continuous control method based on the differential evolution algorithm to realize the position–posture control of a planar passive–active– active–active (PAAA) UM. For the second-order nonholonomic UM, the authors in [12] studied the nilpotent approximation and iterative steering method to achieve the stable control of a planar Pendubot. Liu et al. [13] adopted the fuzzy
Data Loading...
