Professional Documents
Culture Documents
Paper Shared Model
Paper Shared Model
Abstract – Load frequency control of electrical power not only matrix T but also other N-1 controllers. Thus, this
networks to balance the demand and supply based on automatic type of decentralized control is unsuitable for large scale
generation control (AGC) has been widely studied for years. As power networks. In [7] and [8], Andreasson et al propose a
the power networks become more complex, we face the essential hierarchical control structure such that the mechanical power
issue: How to design the load frequency control systematically for each local area is obtained by two control layers: the
to assure the stability of the overall system while optimizing the
proportional (P) control of local frequency ωi in the
global/local control performances? To this end, in this paper,
we propose a framework for glocal (global/local) control of load lower-layer, and in the upper-layer the integral (I) control of
frequency. The command to each local governor includes two the average frequency ϖ = (ω1 + ω2 +…+ ωN)/N. According
signals: the local signal through the corresponding area-control to [8], the overall LFC system is stable if the P gains and I
error (ACE) in the lower-layer, and the global signal gains are both positive. However, this proposition is only
distributed from the upper-layer by controlling the average applicable to the simplified power networks that neglecting
frequency aggregation. We will clarify the role-sharing of the the dynamics of turbines and governors. By selecting the
upper and lower layers via two model sets which are shared in same control gains for the heterogeneous local areas, the
both control layers. Finally, we demonstrate the trade-off authors cannot discuss the control performance of two control
between upper and lower performances through the volumes of
layers. Moreover, the control of tie-line power is neglected by
model sets, and verify the effectiveness of the proposed method
by simulations. this strategy. In [10], Shiltz et al propose excellent
integration of AGC and demand response. However, the
I. INTRODUCTION stability of the overall system is examined by a centralized
In the electrical power networks, load frequency control way as follows. Let z is the vector includes the state of the
(LFC) is the very fundamental function to balance the total system, the dynamics of the overall system is z[k+1] =
demand-supply, and maintain the system frequency at the A∙z[k] where matrix A is established from system dynamics
nominal values [1]. After more than four decades, many and the controllers. Considering the large scale power
control strategies has been proposed. Among them, networks, it is quite complex to select the control gains such
centralized control [2] is no longer of interest since it is that A is Schur stable. In summary, we need a systematic
unsuitable for the very complex power system in the future, design procedure that assuring the stability of the overall
especially when various renewable sources are integrated. system, and enabling control performance analysis.
Thus, completely decentralized control [3~6] or Considering the above issues, the inspiration of our works
hierarchically decentralized control [7~11] has been much is based on the glocal (global/local) concept in which the
attracted in control community in recent years. However, a control actions are restricted locally while the purposes are to
systematic method to assure the stability of the overall load attain both the local and global behaviors [12]. Firstly, we
frequency control system is still a challenge issue. The power show that, similarly to the electric vehicles [13], the load
network can be seen as a large scale system of N local areas, frequency dynamics can be modelled hierarchically with
and the most difficult challenge in stability analysis is the global/local generalized plants by aggregation; and the
interconnection between local areas represented by the physical interaction can be treated in the upper-layer. Then,
synchronizing torque coefficient matrix T of size N×N. we propose a hierarchically decentralized LFC including the
Define Ψ(s) = det(IN + diag{Φi (s)}T/s) where Φi (s) is the upper-layer controller Cg and N lower-layer controllers Cl,i
transfer function of ith local subsystem including the local (Fig. 1). In our original works, only one “global/local shared
plant and local controller. In [6], Tan states that the model set” is shared by the upper-layer and lower-layer [14].
decentralized LFC system is stable if Ψ(s) is stable. However, In this paper, this idea is utilized in a new way. We notice that
it is not easy to assure the stability of Ψ(s), especially when N there exist two channels connecting the control layers,
is a big number and uncertainties are introduced. When namely, Channel 1 for the control distribution from the
tuning any ith controller, it is necessary to know the details of upper-layer, and Channel 2 for the physical interaction
between interconnected areas. The upper-layer expects that
B-M. Nguyen and K. Tsumura are with the Department of Information the transfer functions of the local subsystem in Channels 1
Physics and Computing, the University of Tokyo, Japan. (corresponding author
to provide e-mail: binh_minh_nguyen@ipc.i.u-tokyo.ac.jp).
and 2 belong to the model set Mδ1 and Mδ2, respectively. The
S. Hara was with Department of Information Physics and Computing, the model sets including perturbation can be expressed as
University of Tokyo, Japan. He is now with the Research and Development
Initiative, Chuo University, Japan.
PTL
δ
T IN 1
s
B1
B2
BN
413
zg
WS ( s )
ug
Cg ( s)
yg wg Cg s
Upper-layer controller yg ug
rg1 rg 2
(s) vg
zg Gs wg
1N 1N 1TN N
rg1 rg 2 vg
rl 2
WSL (s)
WSL (s) zl
1N 1N 1TN / N
WSL (s)
rl1
Cl,1(s) Fgt ,1(s) F1 ( s )
yl
Cl,2 (s)
ul
Fgt ,2 (s)
F2 ( s )
vl rl1 rl 2 vl
Cl,N (s) Fgt ,N (s) FN ( s )
Ls
wl
Lower-layer controller
wl zl
d
Q1
Q2
Q N ul yl
Cl ,1 s
1 ( s )
Cl ,2 s
2 (s)
Cl , N s
N ( s )
(a) (b)
Fig.2. Representation of LFC system in case of homogeneous synchronizing torque coefficient.
414
zg WS ( s )
ug yg wg
Cg ( s) g2 g2 zg wg zg 2 wg
control distribution channel
2 p
g p
physical interaction channel g1 g1
(s) vg 1 g2 g2 g2 2 g2
2
g1 g1
1 ( s )
g1 g1 g1 2 g1
1
1 1
rg1
M 1o ( s )
zg Rs wg
g2 g2
s s
2 (s)
M 2o (s)
Cg
rg 2 yg ug
If the above errors satisfy ||Δ1,i ||∞ ≤ δ1 and ||Δ2,i ||∞ ≤ δ2 i, Accordingly, the combined perturbation in Fig. 3(d) is
the overall system can be expressed as in Fig. 3(a) where the
channels from rg1 and rg2 to vg can be represented by the
Δ diag 2 1 , 2 , 2 p
nominal models with the perturbation Δ1 and Δ2 satisfying
1 g
N N Since ||Δ1||∞ ≤ δ1, ||Δ2||∞ ≤ δ2, and ||Δp||∞ ≤ ηg, it is
1 1 ki 1i 1 , 2 1 2i 2
N i 1 N i 1 transparent that Δ( s) 2 . We name {Mjo, δj} (j = 1, 2) as
415
where Δ 1
min : Δ, det I 0
V. ILLUSTRATIVE EXAMPLE
Tg ,i s 1 Tt ,i s 1 Tp , i s 1
Upper-layer design:
where Tg,i is the time constant of the governor; Tt,i is the time Since it is very complex to calculate the μ value directly,
constant of the turbine; Kp,i and Tp,i are the gain and time the standard scheme is to approximate μ by its upper-bound
constant of the generator, with tij = 0.5 i ≠ j and other [16], and the design condition can be reduced to
parameters are
di2
1 0
2
Tg ,1 0.08, Tt ,1 0.3, Tp ,1 20, K p ,1 120, R1 2.4, B1 0.4
d 2j 22
ij
i j
Tg ,2 0.072, Tt ,2 0.33, Tp ,2 25, K p ,2 112.5, R2 2.7, B2 0.4
Tg ,3 0.07, Tt ,3 0.35, Tp ,3 20, K p ,3 115, R3 2.5, B3 0.4 where di is the scaling, i from 1 to 3, and d3 can be set to unit
without loss of generality. The optimal global control
In this paper, we consider the integral controllers in both performance can be obtained using D-K algorithm [16].
layers, or Cl,i (s) = Kl,i /s and Cg(s) = Kg/s. The weighting
functions are assumed to be the 1st oders functions WSL(s) = C. Trade-off between Global and Local Performance
1/(s+λl) and WS(s) = 1/(s+λg). In this paper, due to the complexity of the power network,
it is impossible to obtain the analytical expression of the
B. Controller Design control performances as in our previous works [13]. In stead,
Selection of nominal model: we solve the optimization (13) and (14) by numerical
methods at some discrete values of {δ1, δ2}. We define
We select the 5th order nominal models as
l* min{l*,i } , and summary the results in Fig. 4 considering
i
K po s2
K po s Tgo s 1 Tto s 1 Klo the case that δ1 = δ2. We can see that l* is an increasing
M1o ( s) , M 2o ( s)
Lo ( s) Lo ( s) function of δ1 and δ2. In contrast, g* is shown to be the
decreasing functions of both δ1 and δ2. Thus, there exists the
where N = 3 and
trade-off between the global and local control performances.
Lo ( s) s Tgo s 1 Tto s 1 Tpo s 2 s K po Nt
D. Numerical Simulation
K po Qo s 2 Bo Klo s NtKlo We conduct numerical simulation in Matlab/Simulink
with the step load Pl,1 = 0.01, Pl,2 = 0.02, and Pl,3 = 0.03. Two
We select Klo = 0.5 and {Tgo, Tto, Tpo, Kpo, Qo, Bo} as the
tests are performed for comparison. In test A (Fig. 5), each
averages of the corresponding values in three areas.
local area is provided with the local controller Cl,i (s) = Kl,i /s.
Lower-layer design: In test B (Fig. 6), the secondary control action is generated by
both Cl,i (s) = Kl,i /s and Cg(s) = Kg/s. To design the control
Substitute Φ1,i (s), Φ2,i (s) and the nominal models to the
gains, we select the size of the model sets as {δ1 = δ2 = 0.5}
model matching condition (8) and replace “s” by “jω”, the
which almost balances the global and local control
available set of the local control gain can be expressed as
performances. Thanks to the global control signal based on
Kl ,i Kl ,i : g1,i () 12 & g2,i () 22 0 . On the average aggregation, the deviations of frequencies and
other hand, substitute Sl,i (s) and WSL(s) to (9), the local tie-line powers in Test B attain faster consensus to zero with
control performance must satisfyl2,i fi 0 . Due to less vibration. The hierarchically decentralized control
strategy, therefore, can not only assure the stability of the
the limitation of paper space, we neglect to show the details of
overall system, but also improve the performance of load
g1,i (ω), g2,i (ω), and fi (ω) in this paper. The optimal local
frequency control.
control performance is obtained as
416
(a) Frequency deviation (b) Tie-line power deviation
Fig. 5. Test A: LFC by completely decentralized control.
417