Neural net backlash compensation with Hebbian tuning using dynamic inversion (Q5943845)

From MaRDI portal
scientific article; zbMATH DE number 1648658
Language Label Description Also known as
English
Neural net backlash compensation with Hebbian tuning using dynamic inversion
scientific article; zbMATH DE number 1648658

    Statements

    Neural net backlash compensation with Hebbian tuning using dynamic inversion (English)
    0 references
    0 references
    0 references
    0 references
    8 September 2002
    0 references
    The feedback scheme of actuators backlash compensation is presented. The backlash has piecewise continuous dynamic. The controlled plant is modelled by a system of differential equations written in the Brunovsky canonical form. It is assumed that the unknown nonlinear part is estimated by a bounded function. A robust compensation law is synthesized for the control tracking problem. It is shown that such a backstepping controller guarantees uniformly ultimately bounded tracking error. The required unknown backlash dynamic is approximated with the help of a neural net (NN) technique. The tuning differential law for NN weights based on the Hebbian algorithm is presented. Full stability analysis of the closed loop system is carried out. The computer simulation of the proposed scheme was done for the planar motion of a manipulator link controlled by torque and using backlash.
    0 references
    neurocontrol
    0 references
    backlash compensation
    0 references
    actuator nonlinearity control
    0 references
    dynamic inversion
    0 references
    tracking
    0 references
    neural net
    0 references
    planar manipulator
    0 references

    Identifiers

    0 references
    0 references
    0 references
    0 references
    0 references