Page 295 - 2024-Vol20-Issue2
P. 295

291 |                                                                                                                             Abdulabbas & Salih

                                                                    roe - ro = r˜o,and Ce - C = C˜ yield

        2         I*o  rie  1 - ?2CeLoe  + roe 1 - ?2CeLie  +       Li x? 1  =  -rix1  +  ? Li x2  -    x5  +  Vdc      ?sd    +  L˜ i  d   (iid*)    +  r˜iiid*-
Sd = Vdc                                                                                                        2                       dt

Vg 1 - ?2CeLie                                                      ?L˜ iiiq*

                                                            (30)    Li x? 2  =  -? Li x1    - rix2  -       +  Vdc              +  L˜ i  d      iiq*  + r˜iiiq*+
                                                                                                                2                        dt
                                                                                                        x6              ?sq

                                                                    ?L˜ iiid*

        2    Io*  rie?Ceroe + ?Loe + ?Lie  1 - ?2CeLoe      +       Lo x? 3  =  -rox3  +    ? Lo x4     +  x5  +  L˜ o  d    (iod*)      +   r˜oiod*  -  ?L˜ oioq
Sq = Vdc                                                                                                                dt

? Ce rie Vg                                                         Lo x? 4  =  -? Lo x3    -  rox4  + x6   +     L˜ o  d      ioq*     + ˜roioq* + ?L˜ oiod*
                                                                                                                        dt
                                                            (31)

                                                                    Cx?5 = x1 - x3 + ?Cx6

Equations (30) and (31) represented the switching function          Cx?6 = x2 - x4 - ?Cx5
in steady-state then to find the perturbation state must use the
Lyapunov function. The state variables are clearly defined                                                                                                  (32)
as x1 = iid - iid* , x2 = iiq - iiq*,x3 = iod - iod* = iod - I*o,
x4 = ioq - ioq* = ioq , x5 = vcd - vcd*, x6 = vcq - vcq*.           A Lyapunov function with energy-like properties can be con-
                                                                    structed for this system in the following manner.
5) Lyapunov-function based control
The dynamics of a system near its equilibrium point are often       V(x)        =   ?ELi    + ?ELo      + ?EC     =     3      Lix12     +   3  Lix22+
determined using Lyapunov’s direct technique. The equilib-                                                              2                    2
rium points of a three-phase grid-tied LCL-filtered qZSI is at                                                                                              (33)
(x1 = 0 ,x2 = 0, x3 = 0, x4 = 0, x5 = 0, x6 = 0). This study        3                  3                3                   3
aims to establish a control approach that guarantees the over-      2        Lox32 +   2    Lox42 +     2   Cx5   2  +      2  Cx6   2
all asymptotic stability of a grid-tied qZSI at its equilibrium
point. The state variables converge to the equilibrium point as     It is evident that the first three features described above are met
the total energy of qZSI is continuously dissipated. The qZSI       in Eq. (33). To fulfil the fourth requirement, the function V? (x)
receives its input energy (Ein) from the output voltage (Vdc)
of the qZS network. The energy expended by the components           needs to be determined. Now, differentiating the function v(x)
ri, ro, R, and transistors of the qZSI system results in a loss of
a portion of Ein. Nevertheless, the excess energy is conveyed       and substituting Eq. (32) yield
to the grid, denoted as Eg.
Capacitors and inductors can store energy instead of releasing      V? (x)   =  3     Vdc?sd   +    3       Vdc?sq      - 3ri        x12 + x22           - 3ro  x32+
it. Thus, the energy contained within the qZSI is allocated                     2 x1                2 x2
among the components of the LCL filter, namely the inductor
(Li), capacitor (C), and output filter (Lo). This implies that a    x42 + 3L˜ ix1      d    (iid*)   -  ?   iiq*  + 3L˜ ix2             d    iiq* + ?iid*          +
portion of Ein is transferred through the interchange of energy                        dt                                               dt
stored in the components (?ELi, ?ELo, and ?EC) in a bidirec-
tional manner until the overall energy dissipation approaches       3L˜ ox3     d   (iod*)  -  ?  ioq*  + 3L˜ ox4           d     ioq*       + ?iod*     +
the equilibrium point of the qZSI.                                              dt                                          dt
The control strategy ensuring the overall stability of the grid-
tied qZSI is determined by employing Lyapunov’s direct tech-        3r˜i x1iid* + x2iiq* + 3r˜o x3iod* + x4ioq*
nique. Lyapunov’s direct technique uses a scalar energy-like
function known as the Lyapunov function, represented by V                                                                                                   (34)
(x). According to this approach, a system can be ensured to
be asymptotically stable globally if the Lyapunov function          Assuming perfect matching between actual and estimation
fulfils several requirements such as V(x) = 0, V(x) > 0 at all      parameters the Eq. (34) can be modified by following
x =? 0, V(x) ? 0 as |x| ? 8 ,V? (x) < 0 at all x =? 0. Substi-
tuting the state variables and Eqs. (20)-(25) into Eqs. (14)-       V? (x)   =  3      Vdc?sd +      3      Vdc?sq - 3ri                x12 + x22        -  (35)
(19) and assume Lie - Li = L˜ i , Loe - Lo = L˜ o, rie - ri = r˜i,              2 x1                 2 x2

                                                                    3ro x32 + x42

                                                                    It is apparent that the fourth feature, which is the negative
                                                                    definiteness of V(x), would always be ensured if ?sd and ?sq
                                                                    were selected as

                                                                             ?sd = KdVdcx1                                                                  (36)
   290   291   292   293   294   295   296   297   298   299   300