Professional Documents
Culture Documents
1, March 2020
Abstract: The stability of sampled-data systems is investigated using a new type of Lyapunov functional. The interval from the
sampling point tk to tk +1 is assumed to be a sampling interval. By fully utilizing the characteristic information on the whole
sampling interval, a new two-sided closed-loop Lyapunov functional is proposed, which utilizes the information on both the intervals
from the sampling point t to tk and from t to tk +1 . Based on the two-sided closed-loop Lyapunov functional and modified
free-matrix-based inequality, a less conservative stability criterion is derived for a sampled-data control system, and three numerical
examples are provided to verify the effectiveness and reduced conservativeness of the proposed method. Furthermore, the proposed
method is applied to solve the stability problem of electric power markets, and the practical significance of reducing the
conservativeness is discussed.
Keywords: Stability, sampled-data control system, two-sided closed-loop Lyapunov functional, electric power market
62 Chinese Journal of Electrical Engineering, Vol.6, No.1, March 2020
delay, so that the sampled-data control system was functional, for reducing the conservativeness by
modeled as a linear time-varying time-delay system. effectively mining future sampling information.
The Lyapunov functional and integral inequality The theoretical research on sampled-data control
techniques were used to analyze and design the systems has attracted extensive attention [1-6, 10-11, 13-19].
transformed system. This method overcomes the In the above research, the conservativeness of the
difficulty of dealing with uncertain and variable-period obtained criterion was usually compared by simple
systems faced by the discretization method. numerical examples, which is difficult to use to
Currently, the input time-delay method is widely explain the practical significance of reducing conser-
used to deal with various types of sampled-data vativeness effectively. If an actual system is used to
control problems. For example, the stability and demonstrate the conservativeness of the obtained
stabilization of sampled-data control systems with data criterion, it will be more helpful to explain the
lost was discussed in Ref. [10]. In Ref. [11], the practical significance of reducing conservativeness.
sampling synchronization control problem of neural Therefore, this study takes the electric power market
networks was analyzed. In Ref. [12], the H ∞ output as a practical example to demonstrate the significance
tracking control problem of a T-S fuzzy network of reducing conservativeness, with the aim of
control system was researched. In early studies, the maintaining the supply and demand balance of electric
sawtooth structure features of a sampled-data control power in the grid. Considering the market clearing
system were not fully considered, and therefore the time (i.e., the time of updating power price), the
conclusions obtained were still largely conservative. electric power market is modeled as a linear sampled-
To overcome this disadvantage, some researchers have data control system, and then the conservativeness of
proposed new types of Lyapunov functional, based on the obtained criterion is discussed.
the input time-delay method, such as time-dependent For variable-period sampled-data control systems,
Lyapunov functional[13], time-dependent discontinuous this study proposes a new stability analysis method
Lyapunov functional[14-15] and closed-loop function and applies it to the electric power market. Compared
methods[16-18]. These methods lower conservativeness with the existing research, the main contributions of
as they considered the sawtooth structure features. this study are summarized as follows:
Recently, a time-dependent discontinuous Lyapunov (1) The information of the whole sampling
functional by modified free-matrix integral inequality[19-21] interval is fully utilized. Further, a two-sided closed-
was established in Ref. [19], and the stability problem loop Lyapunov functional is proposed, which not only
of sampled-data control systems was discussed, considers the information of the intervals from the
concluding with less conservativeness. However, only the current state x(t ) to the previous state x(tk ) and
information of the interval between the current sampling from the current state x(t ) to the future state x(tk +1 ) ,
point t and a previous sampling point tk was but also considers the state information of one and
considered, and the information of the interval between the double integral terms and the intervals of [tk , t ) and
current sampling point t and a future sampling point [t , tk +1 ) . Additionally, the cross-information of the
tk +1 was ignored. Additionally, in Ref. [22], the state states mentioned, is incorporated in the two-sided
information of a future sampling point was integrated closed-loop Lyapunov functional. By using the two-sided
into the closed-loop function for the first time, and a closed-loop Lyapunov functional and the modified
new closed-loop function was proposed that free-matrix inequality, the stability criterion of the
effectively reduced the conservativeness of the stable sampled-data control system is derived. The advantages
results of the sampled-data control system. However, and low conservativeness of the proposed method are
this study still ignored a large amount of future verified by two examples.
sampling information, and needs to be further (2) The proposed method is employed to solve
improved. Simultaneously, it also had some conser- the stability problem of the electric power market. This
vativeness to the processing of functional derivative. study not only investigates the impact of clearing time
Therefore, this paper aims to construct a new Lyapunov on power system stability but also discusses the
Tian Zhang et al.: Stability Analysis of Sampled-data Control System and Its Application to Electric Power Market 63
64 Chinese Journal of Electrical Engineering, Vol.6, No.1, March 2020
Tian Zhang et al.: Stability Analysis of Sampled-data Control System and Its Application to Electric Power Market 65
{
Π1 = sym e1T Pe2 + Ξ1T QΞ4 + Ξ T2 QΞ 3 − Ξ1T Z 2 Ξ5 +
.
V C 4 (t ) = η2T (t ) [ Z 3η2 (t ) + 2 Z 4η3 (tk ) ] +
Ξ 3T Z 4 Ξ5 + G1Ξ 6 + D1Ξ 7 − Y1T Ξ 9 − Y2T Ξ10 + (t − tk ){2η2T (t ) [ Z 3η2 (t ) + Z 4η3 (tk ) ] +
N T Γ} − Ξ1T Z1Ξ1 + Ξ 3T Z 3Ξ 3 tk +1
x T (t ) R2 x (t )} − ∫ x T ( s) R2 x ( s )ds =
{ }
t
Π 2 = sym Ξ T2 [ Z1Ξ1 + Z 2Ξ 5 ] + D 2Ξ 7 + Y2T Ξ8 +
ξ T (t ){Ξ 3T ( Z3Ξ3 + 2 Z 4Ξ5 ) + (t − tk ) ×
Ξ 5T M Ξ 5 + e2T R1e2
[2Ξ T4 ( Z 3Ξ 3 + Z 4 Ξ5 ) + e2T R2e2 ]}ξ (t ) −
{
Π 3 = sym Ξ T4 [ Z 3Ξ3 + Z 4Ξ 5 ] + G 2 Ξ 6 + Y1T Ξ8 − } ∫
tk +1
x T ( s ) R2 x ( s )ds (17)
t
Ξ 5T M Ξ 5 + e2T R2 e2
The integral term of the functional derivative
.
G1 = [G1T , −G1T , −2G2T ]
V C 3 (t ) is processed by the modified free-matrix
D1 = [ D1T , − D1T , −2 D2T ] integral inequality in Lemma 1. Then
t
G 2 = [G2T , G2T ,0] D 2 = [ D2T , D2T ,0] − ∫ tk
x T ( s ) R1 x ( s )ds ≤
{
(tk +1 − t ) 2η1T (t ) [ Z1η1 (t ) + Z 2η3 (tk ) ] + Thus, there is a matrix Y1 ,Y2 of any suitable
t dimension that satisfy
x T (t ) R1 x (t )} − ∫ tk
x T ( s ) R1 x ( s )ds = 2ξ T (t )Y1T [ (t − tk )Ξ8 − Ξ 9 ]ξ (t ) = 0 (22)
{
ξ T (t ) −Ξ1T ( Z1Ξ1 + 2 Z 2 Ξ5 ) + (tk +1 − t ) × 2ξ T (t )Y2T [ (tk +1 − t )Ξ8 − Ξ10 ]ξ (t ) = 0 (23)
[2Ξ T2 ( Z1Ξ1 + Z 2Ξ5 ) + e2T R1e2 ]}ξ (t ) − Additionally, since Γξ (t ) = 0 , there is a matrix
t N of suitable dimension that satisfies
∫ tk
x T ( s) R1 x ( s)ds (16)
ξ T (t )( N T Γ + Γ T N )ξ (t ) = 0 (24)
66 Chinese Journal of Electrical Engineering, Vol.6, No.1, March 2020
Summarizing formulas (10) to (24), For the stable sampled-data control system, when
⎡ (t − t ) (t − tk ) ⎤ h1 = 10−5 , according to the stability criterion in
V (t ) = ξ T (t ) ⎢ k +1 Θ1 (tk ) + Θ2 (tk ) ⎥ ξ (t ) (25)
⎣ hk hk ⎦ Theorem 1 and Refs. [16-17, 19, 22], the upper bound
h2 of the maximum sampling period is listed in Tab. 1.
where
It can be seen that, Theorem 1 obtains a larger
Θ1 (tk ) = Π1 + hk Π 2 + hk Δ1
sampling period compared with the results obtained in
Θ 2 (tk ) = Π1 + hk Π 3 + hk Δ 2 Refs. [16-17, 19, 22], indicating that the proposed
Moreover, method has a lower conservativeness, also verifying
h22 T −1 the effectiveness of Note 1.
Δ1 = D1T R2−1 D1 + D2 R2 D2
3 Tab. 1 The maximum allowable upper bound of h2
h2 when h1 = 10−5 for aperiodic sampling
Δ 2 = G1T R1−1G1 + 2 G2T R1−1G2
3 Methods h2
From Schur's supplementary lemma, it is known Ref. [16] 2.515 6
that Θ1 (tk ) < 0 and Θ2 (tk ) < 0 are equivalent to Ref. [17] 2.515 6
linear matrix inequalities in formulas (11) and (12), Ref. [22] (Y1,2 = 0) 2.638 1
Ref. [19] 2.855 4
respectively. If the conditions of formulas (11) and (12)
Theorem 1 (Y1,2 = 0) 3.062 0
are satisfied, then, according to Lemma 1, the sampled- Theorem 1 (Y1 = 0) 3.098 1
data control system in formula (4) is asymptotically Theorem 1 3.109 2
stable, and the proof is completed.
Additionally, in Theorem 1, setting Y1 = 0 ,
Note 3: In the process of deriving Theorem 1,
Y2 = 0 or Y2 = 0 , respectively (that is, to eliminate
the sampled-data control system in formula (4) is used
the effect of formulas (22) or (23), respectively), and
to perform integral operations on the intervals of
the upper bound h2 of the maximum sampling
[tk,t ) and [t,tk +1 ) . The integral states of α1 (t )
interval for the stable sampled-data control system in
and α 2 (t ) are introduced in formulas (22) and (23),
formula (4) is calculated, as shown in Tab. 1. Tab. 1
respectively. Its connection with the system states is
shows that, formulas (18), and (19) are helpful in
established by the free weight matrices of Y1 and Y2 ,
reducing the conservativeness of the system, verifying
that is, more system information is used. The
the effectiveness of Note 3.
zero-value formulas (22) and (23), thus, help reduce
Example 2: The sampled-data control system in
the conservativeness of the system further.
formula (4) is assumed to satisfy the following matrix
Note 4: Theorem 1 provides a stability criterion
parameters
for a variable-period sampled-data control system.
⎡ 0 1⎤ ⎡0 0⎤
Obviously, if h1 = h2 in Theorem 1, the same A=⎢ ⎥ Bc = ⎢
⎣ −2 0.1⎦ ⎣1 0 ⎥⎦
criterion applies to the constant-period sampled-data
For the stable sampled-data control system in
control system. Additionally, Theorem 1 is also
formula (4), when h1 has different values, the
applicable to the sampled-data control systems of A
maximum sampling interval is calculated using
and Bc with polyhedral uncertainties through
Theorem 1 and Refs. [6, 8, 17-19]. The results are
appropriate transformations.
shown in Tab. 2.
4.2 Example simulation
Tab. 2 Intervals of allowable aperiodic samplings [ h1 , h2 ]
This subsection illustrates the effectiveness of the Methods Sampling intervals [h1 , h2 ] Sampling intervals [h1 , h2 ]
proposed method through two numerical examples. Ref. [17] [0.4, 1.31] [0.8, 1.56]
Example 1: The sampled-data control system in Ref. [6] [0.4, 1.39] [0.8, 1.61]
formula (5) is assumed to satisfy the following matrix Ref. [8] [0.4, 1.43] [0.8, 1.58]
Tian Zhang et al.: Stability Analysis of Sampled-data Control System and Its Application to Electric Power Market 67
It can be seen from Tab. 2 that the proposed coefficient, and E is the integral of the difference
method can obtain a larger sampling interval compared between the power supply and the demand.
with [6,8,17-19], indicating that the proposed method (2) Market consumption model.
is better and less conservative, further verifying the The marginal benefit and the demand for
effectiveness of Note 1. electrical power satisfy the following relationship
λd = bd + cd Pd (28)
5 Application to electric power markets
where, λd is the marginal income, bd is the fixed
A prerequisite for the reliable and stable
income, cd is the fixed coefficient, and Pd is the power
operation of the grid is that the output and the load of
demand. When marginal returns are higher than marginal
the electric power (i.e., supply and demand) are
prices, consumers will expand their consump- tion, and the
balanced at all times. To maintain power balance, the
speed of expansion is related to consumers. Therefore, the
scheme of load frequency control is usually adopted [26].
market consumption model can be described as
In recent years, researchers have proposed that the
τ d Pd = bd + cd Pd − λ (29)
electric power market can use power prices to
maintain the balance between supply and demand, where, τ d is the time constant representing the
which has attracted extensive attention [15, 27-29]. This expansion speed of consumer demand.
section applies the proposed method to the stability (3) Unbalanced power supply and demand as well
analysis of the electric power market (mainly as market price response.
including modeling the electric power market) and The imbalance between the supply and the
theoretical calculation (for example simulation and demand can be expressed by the integral of the
discussion). difference between the power supply and demand as
E = Pg − Pd (30)
5.1 Models of the electric power market
The change in the power price on the market is
A simple electric power market consists of power determined by observing the demand and supply of the
suppliers, power users, and real-time market. grid. This is given by
(1) Market supply model. τ λ λ = − E (31)
The marginal production cost of power generation
where, τ λ is the time constant representing the
and the power supply satisfy the following relationship
response speed to the power price at the time of
λg = bg + cg Pg (26)
market disturbance.
Here, λg is the marginal cost, bg is the fixed cost of In the actual electric power market, the discrete power
the supplier, cg is the fixed coefficient, and Pg is price signal received at a certain interval is equivalent to
the amount of power supplied. If the power supplier the market-clearing time, meaning the period of updating
observes that the power price on the market is higher the power price [28]. According to formula (27) and
than its production cost, the power supplier will formulas (29) to (31), the following linear control model
expand production, and the degree of expansion will based on sampled-data control is obtained.
be proportional to the difference between the observed ⎧τ g Pg (t ) = λ (tk ) − bg − cg Pg (t ) − kE (t )
⎪
market power price and the actual production cost. ⎪τ d Pd (t ) = bd + cd Pd (t ) − λ (tk )
⎨ (32)
Furthermore, additional costs will be incurred based ⎪ E (t ) = Pg (t ) − Pd (t )
on an oversupply of power at some point in the past. ⎪τ λ λ (t ) = − E (t )
⎩
Therefore, the model of the electric power market Here, tk is the clearing time for the power price
supply can be described as and it satisfies
τ g Pg = λ − bg − cg Pg − kE (27) tk +1 − tk = Tmct Tmct ∈ [Tmct , Tmct ]
k k m
(33) M
where, τ g is the time constant representing the where, Tmctk and TmctM represent the update period
response speed of the power output, λ is the of the kth market power price and its maximum value,
observed power price on the market, k is the fixed respectively.
68 Chinese Journal of Electrical Engineering, Vol.6, No.1, March 2020
⎡ b bd ⎤
J = ⎢− g 0 0⎥
⎣ τg τd ⎦ Additionally, when the initial conditions are set as
⎡ cg ⎤ y (t ) = [20,20,0,0]T , and the sampling period of the
⎢− τ 0 −k 0⎥
τg electric power market price signal is Tmct = 10.64 , the
⎢ g ⎥
⎢ cd ⎥ stable balance point of the system in formula (34) is
0 0 0⎥
A=⎢ τd given by
⎢ ⎥
⎢ 1 −1 0 0⎥ y* = [ Pg* , Pd* , E * , λ * ]T = [26.68, 26.68,0, 4.66]T
⎢ ⎥
⎢ 0 0 − 1 0⎥ The state trajectory of the system is shown in Fig.
⎢⎣ τλ ⎥⎦
1. The figure shows that when the update period of the
⎡0 0 0 1 ⎤
⎢ τg ⎥ price signal is Tmct = 10.64 , the system is stable. The
⎢ ⎥ discrete control signal going through the zero-order
⎢
Bc = ⎢0 0 0 −1⎥
τd ⎥ holder (ZOH) is shown in Fig. 2.
⎢ ⎥
⎢0 0 0 0 ⎥
⎢⎣0 0 0 0 ⎥⎦
Assuming that y* = [ y1* , y2* ,", yn* ] is the balance
point of the system in formula (34), to facilitate the
processing, the equilibrium point y* is moved to the
origin through coordinate transformation, and then the
system in formula (34) can be transformed into
x (t ) = Ax(t ) + Bc x(tk ) (35)
This system has been discussed in Section 4. The
key parameters A and Bc of the system in formula
(35) have the same physical meanings as those defined
by the system in formula (34). Fig. 1 State trajectories of the system (34) during
Therefore, the proposed stability analysis method Tmct = 10.64
of the sampled-data control system can also be used to
discuss the stability of the electric power market.
Tian Zhang et al.: Stability Analysis of Sampled-data Control System and Its Application to Electric Power Market 69
70 Chinese Journal of Electrical Engineering, Vol.6, No.1, March 2020
model. Acta Automatica Sinica, 2015, 41(3): 661-668. [24] S B Ding, Z S Wang, N N Rong, et al. Exponential
[13] E Fridman. A refined input delay approach to sampled- stabilization of memristive neural networks via saturating
data control. Automatica, 2012, 46(2): 421-427. sampled-data control. IEEE Transactions on Cybernetics,
[14] K Liu, E Fridman. Wirtinger’s inequality and Lyapunov- 2017, 47(10): 3027-3039.
based sampled-data stabilization. Automatica, 2012, 48(1): [25] G Guo, L Ding, Q L Han. A distributed event-triggered
102-108. transmission strategy for sampled-data consensus of
[15] C K Zhang, Y He, L Jing, et al. Stability analysis of multi-agent systems. Automatica, 2014, 50(5): 1489-1496.
sampled- data systems considering time delays and its [26] L Jing, W Yao, J Y Wen, et al. Delay-dependent stability
application to electric power markets. Journal of the for load frequency control with constant and time-varying
Franklin Institute, 2014, 351(9): 4457-4478. delays. IEEE Transactions on Power Systems, 2012, 27(2):
[16] A Seuret. A novel stability analysis of linear systems 932-941.
under asyn-chronous samplings. Automatica, 2012, 48(1): [27] F L Alvarado. The stability of power system market.
177-182. IEEE Transactions on Power Systems, 1999(142):
[17] A Seuret, F Couaisbaut. Wirtinger-based integral inequality: 505-511.
Application to time-delay systems. Automatica, 2013(49): [28] J Nutaro, V Proropopescu. The impact of market clearing
2860-2866. time and price signal delay on the stability of electric
[18] A Seuret, C Briat. Stability analysis of uncertain power markets. IEEE Transactions on Power Systems,
sampled-data systems with incremental delay using 2009, 24(3): 1337-1345.
looped-functionals. Automatica, 2015(55): 274-278. [29] L Shen, H Q Xiao. Delay-dependent robust stability
[19] T H Lee, H P Ju. Stability analysis of sampled-data analysis of power systems with PID controller. Chinese
systems via free-matrix-based time-dependent discontinuous Journal of Electrical Engineering, 2019, 5(2): 79-86.
Lyapunov approach. IEEE Transactions on Automatic
Control, 2017, 48(1): 3653-3657. Tian Zhang received his B.S. degree in electrical
engineering from Hunan University of Technology,
[20] H B Zeng, Y He, M Wu. Free-matrix-based integral Zhuzhou, China, from September 2014 to July
inequality for stability analysis of systems with time-varying 2018. He is currently a postgraduate student at
Hunan University of Technology. His current
delay. IEEE Transactions on Automatic Control, 2015,
research interest is robust control in power
62(7): 2768-2772. systems.
[21] H B Zeng, K L Teo, H Xu, et al. Sampled-data synchro-
nization control for chaotic neural networks subject to Shenping Xiao received the B.S. degree In
actuator saturation. Neurocomputing, 2017, 18(6): 25-31. automation from Northeastern University,
Shenyang, China, in 1988, the M.S. degree in
[22] H B Zeng, K L Teo, Y He. A new looped-functional for
computer science from the Central South
stability analysis of sampled-data systems. Automatica, University of Forestry and Technology, Changsha,
2017(82): 328-331. China, in 2002, and the Ph.D. degree in control
theory and control engineering from Central
[23] Z S Wang, J Sun, H G Zhang. Stability analysis of T-S
South University, Changsha, in 2008. He has
fuzzy control system with sampled-dropouts based on been with the Department of Electrical and
time-varying Lyapunov function method. IEEE Information Engineering, Hunan University of Technology, Zhuzhou,
China, where he is currently a Professor of automatic control
Transactions on Systems, Man, and Cybernetics: Systems, engineering. His current research interests are time-delay systems,
DOI: 10.1109/TSM- C.2018.2822482. networked control systems, and robust control in power systems.