s9p7

Embed Size (px)

Citation preview

  • 8/8/2019 s9p7

    1/7

    Synergetic Synthesis Of Dc-Dc Boost Converter Controllers:

    Theory And Experimental Analysis

    A. Kolesnikov (+), G. Veselov (+), A. Kolesnikov (+), A. Monti (++), F. Ponci (++), E. Santi (++), and

    R. Dougal (++

    )

    (+

    )Department of Automatic Control SystemTaganrog State University of Radio-Engineering (TSURE)

    44 Nekrasovsky St., Taganrog, 347928, Russia

    (++

    )Department of Electrical Engineering

    University of South Carolina

    Swearingen Center, Columbia, SC 29208 U.S.A.

    Abstract- This paper describes a new approach to the

    synthesis of controllers for power converters based on the

    theory of synergetic control. The controller synthesis

    procedure is completely analytical, and is based on fully

    nonlinear models of the converter. Synergetic controllers

    provide asymptotic stability with respect to the required

    operating modes, invariance to load variations, and

    robustness to variation of converter parameters. With respectto their dynamic characteristics, synergetic controllers are

    superior to the existing types of PI controllers. We present

    here the theory of the approach, a synthesis example for a

    boost converter, simulation results, and experimental results.

    I. INTRODUCTION

    Design of controllers for power converter systems presents

    interesting challenges. In the context of system theory,

    power converters are non-linear time-varying systems; they

    represent the worst condition for control design.

    Much effort has been spent to define small-signal linear

    approximations of power cells so that classical control

    theory could be applied to the design. See for example

    [1,2]. Those approaches guarantee the possibility to use a

    simple linear controller, e.g. Proportional-Integral

    controller, to stabilize the system.

    The most critical disadvantage is that the so-determined

    control is suited only for operation near a specific

    operating point. Further analyses are then necessary to

    determine the response characteristics under large signal

    variations [3,4].

    Other design approaches try to overcome the problem by

    using the intrinsic non-linearity and time variation for the

    control purpose. Significant examples of this approach

    include the sliding mode control, used mostly for

    continuous-time systems [5] and the deadbeat control, used

    for digital systems [6]. Those two theories have beenapplied to power electronics mostly because of their

    intrinsic capability to manage variably-structured systems.

    In this paper we focus on a different approach, synergetic

    control [7], that tries to overcome the previously described

    problems by using the internal dynamic characteristics of

    the system.

    The synergetic approach is not limited by any non-

    linearity; instead, it capitalizes on such non-linearities.

    As will be discussed in the paper, this approach makes full

    use of the intrinsic proprieties of the system. While this is a

    strong point, it is also a weak point -- definition of the

    system model plays a more strategic role than in any other

    control approach. This introduces a great possibility for

    sensitivity to system parameters. However, as we will

    demonstrate with experimental results, this problem can besolved.

    One obvious solution is the adoption of sophisticated

    observers for parameter determination. This solution is

    reasonable only if the cost of the control is not a significant

    concern (e.g. high-power or high voltage applications). For

    situations where the control costs are of concern, we will

    show that suitable selection of the control macro-variables

    can largely resolve any sensitivity to uncertainty in system

    parameters.

    In this paper we will describe the theory of synergetic

    control, demonstrate its application in the case of a boost

    converter, describe both simulation and experimental

    results, and finally introduce some interesting practical

    considerations.

    II. THEORETICAL BACKGROUND

    Synthesis of a synergetic controller begins by defining a

    macro-variable, which is a function of the system state

    variables:

    ),()( txt = (1)The control objective is to force the system to operate on

    the manifold 0= . The designer can select thecharacteristics of this macro-variable according to the

    control specifications (e.g. limitation in the control output,

    and so on). In the trivial case is a simple linear

    combination of the state variables. This process is thenrepeated, defining as many macro-variables as there are

    control channels

    Next, the dynamic evolution of the macro-variables is

    fixed according to the equation:

    ( ) 0;0 >=+ TtT (2)where T is a design parameter describing the speed of

    convergence to the manifold specified by the macro-

  • 8/8/2019 s9p7

    2/7

    variable. Finally, the control law (evolution in time of the

    control output) is synthesized according to equation (2) and

    the dynamic model of the system.

    Briefly, any manifold introduces a new constraint on the

    domain of the state space, and thereby reduces the order of

    the system and forces it in the direction of global stability.

    The procedure summarized above can be easily

    implemented as a computer program for automatic

    synthesis of the control law or it can be performed by hand

    for simple systems, such as for the boost converter, that

    have a small number of state variables.

    By suitable selection of macro-variables the designer can

    obtain interesting characteristics for the final system such

    as:

    Global stability Parameter insensitivity Noise suppression

    These results are obtained while working on the full

    nonlinear system and the designer does not need to

    introduce simplifications in the modeling process to obtain

    a linear description as is required for classical control

    theory.

    III. SYNTHESIS OF A SYNERGETIC CONTROLLER FOR A BOOST

    CONVERTER

    We now synthesize a controller for a DC-DC boost

    converter (see Fig. 1). The classical time-averaged model

    of the converter is:

    ( ) ( )

    ( ) ( ) ,1

    ;1

    1

    212

    21

    RC

    xu

    C

    xtx

    VL

    uL

    xtx g

    =

    +=

    C

    C

    (3)

    10 u (4)where 1x is the inductor current, 2x the capacitor voltage

    and u the switch duty cycle. Our objective is to obtain the

    control law ( )21,xxu as a function of state co-ordinates 1x ,

    2x , which provides the required values of converter output

    voltage sxx 22 = and, therefore, current Sxx 11 = forvarious operating modes, while satisfied limitation (4).

    Fig. 1: Boost Converter scheme

    According to this method, we introduce the following

    macro variable

    0; 12111 >= xx (5)

    Substitution of 1 (3) into the functional equation:

    ( ) 0;0 1111 >=+ TtT (6)yields:

    ( ) ( ) 01

    1

    1

    211 =+ T

    txtx (7)

    Now substituting the derivatives ( )tx1 and ( )tx2 from (3)and (4), the control law is obtained:

    +++

    == 121

    112

    1

    111

    TV

    LRC

    x

    LxCx

    LCuU g (8)

    The expression for u1 is the control action for the converter

    controller. Substituting macro variable and RCT = into (8), we obtain the control law as:

    ( )

    ++

    += gV

    Lx

    RCRC

    x

    LxCx

    LCu

    111 2

    11

    112

    1

    (9)

    When 1= , i.e. RCT =1 , we get:

    +

    += gV

    LRC

    x

    LxCx

    LCu

    11 1

    112

    1

    (10)

    Control laws (8), (9), or (10), according to (6), inevitably

    move the representing point (RP) of object (1) firstly to in-

    variant manifold 01 = (3), and then along this manifoldto the converters steady state: sxx 11 = ; sxx 22 = . Let usstudy the behavior of the closed loop system:

    ( )

    ( )RC

    x

    TV

    LRC

    x

    CxLx

    Lxtx

    VLTVLRC

    x

    CxLx

    Cxtx

    g

    gg

    21

    1

    21

    211

    12

    1

    1

    21

    211

    21

    11

    ;111

    ++

    +=

    +

    +++=

    D

    D (11)

    on the manifold 0= (3). For this purpose, we substituterelation 21 xx = into (11). This results in:

    ( )( )

    ( )( )

    .1

    ;

    21

    1

    221

    2

    2

    1

    21

    2

    1

    1

    1

    CL

    Vx

    CLRtx

    CL

    V

    CLR

    xtx

    g

    g

    ++

    +=

    ++

    +=

    D

    D

    (12)

    Each separate equation of (12) describes the behavior ofthe corresponding converter coordinate 1x or 2x on the

    manifold 01 = . Evidently, equation (12) isasymptotically stable with respect to the converters steady

    state:

    gsgs RVxRVx 122

    11 ; == (13)From relation (13) we see that converters steady state

    operating point depends on the power source voltage

  • 8/8/2019 s9p7

    3/7

    gV and on the load resistance R. After we set the required

    reference value of the converters output voltage sx2 , (13)

    gives us a possibility to find 1 , present in macro variable

    1 , i.e.

    g

    s

    g

    s

    RVx

    RVx 121 == . (14)

    The steady state value of control su , which provides the

    converters steady state (13), will be determined by the

    following equation:

    111 =Su . (15)

    Knowing 1 in (14) and Su1 in (15), we can find the

    steady state parameters of the converter.

    So, the synthesized control law 1u after a time interval

    approximately equal to T3 moves the RP of the plant to

    the manifold 01 = , and then, according to equation (12),provides asymptotically stable movement along 01 = tothe converters steady state (13). According to (12), the

    time to move RP along 01 = is determined by the

    expression ( )CLR +21 .

    Control law 1u provides converter motion from an

    arbitrary initial state 010 x , 020 >x to the steady state

    sx1 , sx2 . In other words, the synthesized control law 1u

    with 01 >T and 01 > guarantees asymptotic stability (inthe whole) of the closed loop system with respect to the

    converters steady state.

    IV. OTHER POSSIBLE CONTROL SYNTHESIS FOR THE BOOST CASE

    The previous case illustrated a very simple case of control

    synthesis that transformed the boost circuit into a first

    order system always working in the manifold described by

    the macro-variable. This case does not cover all the

    possible situations we could face in reality, where more

    complex macro-variables must be introduced.

    One classical problem is accounting for limitation of one

    of the state variables, for example, limiting the maximum

    input current. This problem can be simply solved by

    defining a new macro-variable:)tanh( 2212 xAx = (16)

    where max1xA = . This defines a new manifold where thecurrent is naturally limited. In the rest of paper other

    assumptions will bring us to the definition of other possible

    macro-variables.

    V. SYSTEM MODELING RESULTS

    Extensive simulation analysis has been conducted to verify

    the control performance. The simulations have been

    performed using both Matlab and the VTB simulator [8].

    Fig. 2 and Fig. 3 show the transients created by changes in

    the load and in the power source amplitude, as predicted by

    Matlab models. Fig. 4 shows the phase portrait of thesystem and the stability characteristics of the control

    system as demonstrated by convergence to the manifold.

    Other simulation results, obtained in the VTB

    environment, are shown in Fig. 5 and Fig. 6. This second

    step was useful insofar as guiding construction of the real

    converter because of the possibility to use more detailed

    models of the power cell. For example, the capacitance

    model in VTB contains also the equivalent series

    resistance, giving the opportunity to explore more realistic

    problems.

    Fig. 2: The voltage transients

    Fig. 3: The load changing

  • 8/8/2019 s9p7

    4/7

    Fig. 4: System phase portrait

    VI. LABORATORY EXPERIENCE

    Following theoretical analysis, a laboratory prototype was

    designed and built. Since synergetic control is well suited

    for digital implementation a DSP-based platform wasselected for migration from the VTB environment to the

    real world.

    The small-scale power converter system has the following

    nominal characteristics:

    - Rated Input Voltage: 12 V

    - Rated Output Voltage: 40 V

    - Maximum Load: 100 W

    - Input Inductance: L = 46 mH

    - Output Filter Capacitance: 1.360 mF

    - Main Switch: IRF540N

    The main targets of the experimental analysis were:

    Verification of the control theory Analysis of problems related to the model

    parameter sensitivity

    By defining the controller in Simulink, we were able to

    easily export the control algorithms to both a dSpace

    platform for control of the real hardware, and to the VTB

    environment for system simulation. The ease of inserting

    the Simulink controller into both hardware model allowed

    unique opportunities to rapidly experiment with a wide

    variety of macro-variable definitions in order to identify

    and resolve significant early problems.One interesting observation common to both experimental

    environments (simulation and hardware) was the

    possibility to introduce any kind of transient in the output

    voltage reference without requiring any soft-start option.

    The system easily remained stable under large non-linear

    transients.

    Fig. 5: VTB schematic

    Fig. 6: VTB results

    On the other hand, adoption of the simplest macro-variabledefinitions revealed significant problems with respect to

    parameter sensitivity. This sensitivity mostly affected the

    steady-state value of the output voltage -- which resulted to

    be different from the reference value.

    For this reason, after the first set of experiments, a new

    macro-variable was defined:

    )x-(xk)x-(x 1ref12ref2 += (17)

    This new macro-variable significantly reduced the problem

    of parameter sensitivity and allowed for the steady state to

    be set more accurately.

    Using this approach two main parameters had to be tunedfor control performance:

    The value T involved in the main synergeticequation (2)

    The value of K involved in the macro-variabledefinition.

  • 8/8/2019 s9p7

    5/7

    The role of T is extremely interesting. As far as equation

    (2) is concerned, T defines the speed with which we reach

    the manifold. On the other hand, this parameter also plays

    an interesting role in noise reduction.

    In the case of the boost control, the state vector is easily

    accessible and so we can assume that the error introduced

    in evaluation of the macro-variable is quite limited. On the

    other hand, its derivative is obtained by means of the state

    equations so then the system parameters play a significant

    role.

    Let us suppose that we have a systematic constant error in

    the evaluation of the derivative. If we check for the steady-

    state condition of this equation we will have:

    ( ) 0)( =++ etT D

    and then in steady state:

    Te=

    This means that by decreasing T, we decrease the time

    with which the manifold is reached. But also we reduce the

    steady-state error that is introduced by wrong estimation of

    the system parameters. During the experiments we found

    that a reduction of T from 1ms to 0.1 ms yielded a

    significant increase in accuracy of the steady-state.

    K plays a significant role after reaching the manifold; it

    determines the way that errors in the main state variable

    are canceled by using the error on the current.

    Decreasing K increases the control performance but also

    calls for a higher current peak during any transient.

    This situation, as pointed out in the introduction, could be

    solved by definition of a more sophisticated macro-

    variable which can account for current limitation.

    The synergetic approach also gives an opportunity to solve

    the steady-state problem by introducing a new state

    variable that represents the integral of the reference-

    feedback error. This is analogous to an integral term in a

    standard linear controller.

    We decided to avoid this option in order to keep the system

    simpler and to better exploit the possibilities offered by

    parameter tuning and macro-variable definition.

    However, the introduction of the integral term is always

    possible to force the error to go zero at steady state.

    According to the laboratory experience, we also figured

    out that this option should be considered eventually as a

    refining option working in the direction of keeping theintegral charge as small as possible.

    VII. COMPARISON BETWEEN SIMULATION AND EXPERIMENTAL

    RESULTS

    All the results shown in the following have been obtained

    using the macro-variable definition reported in (17).

    We want now to show some comparisons between

    simulated and experimental results that confirm the

    theoretical discussion presented in the previous paragraph.

    These results show the transient that follows step change of

    the reference voltage from 20 to 40 V.

    Fig. 7: Output Voltage (Simulation)

    Fig. 8: Output Voltage (experiment)

    Fig. 7 and Fig. 8 show the output voltage from experiment

    and simulation. One can clearly see that the synergetic

    control transformed the second-order system into a first-

    order system. This can be easily justified by consideringthat when we are on the manifold we have a linear relation

    between two state variables. Introducing the constraint, the

    order is reduced. This is always true for synergetic

    applications and it constitutes a similarity with the sliding

    mode approach.

    Fig. 9 and Fig. 10 focus on the evolution in time of the

    macro-variable. The two transients looks very similar in

    the first part moving in the direction of the manifold with

    the same speed. In the experimental results, anyway, a

  • 8/8/2019 s9p7

    6/7

    second transient starts when we are close to steady state:

    this can be considered another side effect of the imperfect

    system modeling.

    Fig. 9: macro-variable as function of time (simulated data)

    Fig. 10: macro-variable as function of time real data)

    Finally in Fig. 11 and in Fig. 12 the results for the input

    current are presented. Also in this case the simulation

    results and the experimental data match perfectly.

    VIII. CONCLUSIONS

    This paper introduced a new and interesting control

    approach called Synergetic Control. The main feature ofthis approach is to manage with the same level of

    simplicity both linear and non-linear systems.

    The main aspect of control design is definition of a macro-

    variable that specifies a manifold for the space variables.

    We have discussed several different definitions of the

    macro-variable and described the practical consequences of

    the different selections. The theoretical aspects have been

    discussed and then confirmed through experiment and

    simulation.

    Fig. 11: Input Current (averaged-simulated value)

    Fig. 12: Input current (filtered-experimental data)

    ACKNOWLEDGEMENT

    This work was supported by the US Office of Naval

    Research (ONR) under grant N00014-00-1-0131

    REFERENCES

    [1] S. Sanders, J. Noworolski, Xiaojun Z. Liu, and G. C.

    Verghese, Generalized Averaging Method for Power

    Conversion Circuits, in IEEE Trans. on Power

    Electronics, vol. 6. N. 2, April 1991, pp. 251-258[2] D.M. Mitchell, "DC-DC switching regulator analysis",

    McGraw Hill Book Company, 1988

    [3] R. W. Erickson, S. Cuk, and R.D. Middlebrook,

    Large-scale modelling and analysis of switching

    regulators, in IEEE PESC Rec., 1982, pp. 240-250

    [4] P. Maranesi, M. Riva, A. Monti, A. Rampoldi,

    "Automatic Synthesis of Large Signal Models for

  • 8/8/2019 s9p7

    7/7

    Power Electronic Circuits", IEEE-PESC99, Charleston

    (USA), July 1999

    [5] V.I. Utkin, "Variable Structure system with Sliding

    modes". IEEE Trans. on Ind. Electronics, vol AC 22,

    no. 2, pp. 212-222, 1977.

    [6] L.Ben-Brahim, A. Kawamura, Digital Control of

    Induction Motor Current with Deadbeat Response

    Using Predictive State Observer, IEEE Trans. On

    Power Electronics, vol. 7, N. 3, July 1992, pp. 551-559

    [7] A. Kolesnikov, G. Veselov, A. Kolesnikov, et al.

    Modern applied control theory: Synergetic Approach

    in Control Theory, vol. 2. (in Russian) Moscow

    Taganrog, TSURE press, 2000

    [8] R. Dougal, T. Lovett, A. Monti, E. Santi, A

    Multilanguage Environment for Interactive Simulation

    and Development of Controls for Power Electronics,

    IEEE PESC01, Vancouver (Canada).