Nonlinear Feedback linearisation

Embed Size (px)

Citation preview

  • 7/28/2019 Nonlinear Feedback linearisation

    1/17

    A feedback linearizationcontrol the nonlinear 5-Do

    flywheel suspended by PMbiased HMB

    Authors: Wen Tong, Fang Jiancheng

    Presen

    Arpit Ch

    Digvija

  • 7/28/2019 Nonlinear Feedback linearisation

    2/17

    Contents Introduction

  • 7/28/2019 Nonlinear Feedback linearisation

    3/17

    Intro AMBs(Active magnetic bearings) instead of normal ball-bea

    Advantage: High Rotation speed

    No contact character

    Nearly zero friction

    Long life span

    No auxiliary lubricant system

    Types:

    EMB(Electro-magnetic bearing)

    PM(permanent magnet) biased HMB(Hybrid magnetic bearing)

  • 7/28/2019 Nonlinear Feedback linearisation

    4/17

    EMB vs PM biased HMB

    In EMB, suspending forces for rotor are generated completecurrents.

    Whereas in HMBs PM provide the biased flux and current gthe regulating force to realize the stable suspension of the r

    Power consumption is reduced greatly in the application of

    biased HMB, because the biased flux is proved by the permamagnet without any external power consumption as opposebiased flux is generated by biased current.

  • 7/28/2019 Nonlinear Feedback linearisation

    5/17

    Need of controller

    Effective controller is required for Magnetic suspensionflywheel(MSFW) because of the the instability of the open-lAMB caused by the negative displacement stiffness of themagnetic force .

    The dynamic model of the PM biased HMB is nonlinear andcomplicated, and its nonlinearity is caused by the superposi

    PM flux and EM flux and the magnetic flux coupling in differdirections.

    It is simplified by assuming that permanent magnet in HMB equivalent to biased current in the EMB, so an equivalent EMmodel describing the HMB system can be obtained

  • 7/28/2019 Nonlinear Feedback linearisation

    6/17

    Need of exact linearization

    Based on the equivalent model, using the Taylor series expa

    the neighborhood of the operating point(or equilibrium poiapproximate linear model of the nonlinear HMB system canobtained for the controller design.

    This method has an advantage that the control model is verand the controller can be realized simply.

    For the application of the HMB with high restriction on powand accuracy(such as MSFW), the controller design must bethe nonlinear model of system.

  • 7/28/2019 Nonlinear Feedback linearisation

    7/17

    Sketch of MSFW

    Diagram 1:

  • 7/28/2019 Nonlinear Feedback linearisation

    8/17

    Model development For convenience, there are three coordinate frames introdu

    system, as shown in Figure. OXYZ is the stator-fixed frame, o

    rotor-fixed frame and oabc is the intermediate frame.

  • 7/28/2019 Nonlinear Feedback linearisation

    9/17

    Five DOFs According to the design of flywheel(MSFW), The rotation alon

    direction is controlled by the driven motor and independent of tThe translation z is controlled by the axial HMB and can be regardecoupled with other four coupled axial DOFs[x, y, , ].

    The four axial DOFs of the rotor could be calculated from the foupicked up by the displacement sensors[xma, yma, xmb, ymb] as showEq.(1):

    x = xma+ xmb

    = (xma- xmb)/L

    y = yma+ ymb

    = (yma- ymb)/L

    L is the distance between the distance between the two displacesensors located at the two ends the stator of the MSFW, as showfollowing figure

  • 7/28/2019 Nonlinear Feedback linearisation

    10/17

  • 7/28/2019 Nonlinear Feedback linearisation

    11/17

    MSFW description

    The MSFW includes four sets of radial HMBs.

    As shown in following figures, each set consist of two seriesconnected coils placed diagonally opposite to generate restforces.

  • 7/28/2019 Nonlinear Feedback linearisation

    12/17

    rotor

  • 7/28/2019 Nonlinear Feedback linearisation

    13/17

    Take the HMB in X direction for example. It is assumed that the control cur

    sent to the opposite connected coils in X+ and X-. Control currents flowing

    two coils in X+ and X- are iax and iax, respectively. fax is the resultant restor

    acting on the rotor, which is the difference between the two magnetic forc

    in X+ and X- directions.

    fax = fax+ - fax-= [(pax++ iax++ iax-ax+)2 - (pax- - iax- - iax+ax-)

    2 ]/ (20A)

    Where,

    pax+, pax- are PM flux in air-gap of X+ and X-,

    iax+

    , iax+ax-

    are EM flux in X+ and X- generated by the control current iax

    ,

    iax-, iax-ax+ are EM flux in X+ and X- generated by the control current -iax.

    permeability of the vacuum and A is the area of one pole. The other three

    forces fay, fbx and fby would be obtained using the same methods described

  • 7/28/2019 Nonlinear Feedback linearisation

    14/17

    Rotor Dynamics

    ma and mb are distances from HMBs to the origin of the OXYZ.

    fx, fy, px, and py are the disturbances caused by the residual static and dynamic

    imbalances.

    The state variables are chosen as [x y ]. The generalized magnetic forces actin

    on the system are [fx py fy px].

  • 7/28/2019 Nonlinear Feedback linearisation

    15/17

    Dynamic model of the nonlinear sys the MSFW is a 4-input, 4-output, 8-order nonlinear system.

    The dynamics model is consisted of three parts. The first is sequation f(x) which is regarded as the uncontrollable permamagnetic force and the couple caused by the gyroscopic efferotor. The second is the controllable magnetic force gi (x)uithe input part of the system. The inputs of the system are seu=[iax ibx iay iby]. The last one is the disturbances caused by thdynamic imbalances.

  • 7/28/2019 Nonlinear Feedback linearisation

    16/17

    where m is themassoftheroto

    the moments ofinertialofthe

    directions.

  • 7/28/2019 Nonlinear Feedback linearisation

    17/17

    The input functions of the nonlinear system are described in detail as follows:

    The output functions for the selected outputs: