IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS, VOL. 17, NO.

1, JANUARY 2021 219

Nonlinear Mean-Square Power Sharing

Control for AC Microgrids Under
Distributed Event Detection
Jingang Lai , Senior Member, IEEE, and Xiaoqing Lu , Senior Member, IEEE

Abstract—This article proposes a nonlinear distributed as an effective solution. A microgrid is an aggregated entity
cooperative control scheme that can regulate the power that integrates distributed energy resources (DERs), such as
output to achieve efficient utilization of renewable energy photovoltaics (PVs) and battery energy storage systems (BESSs)
in ac microgirds, which ensures mean-square autonomous
proportional power sharing over a nonlinear microgird [1], [2]. The microgrid can operate both connected to the main
system via a sparse cyber network subject to noisy dis- grid or in islanded mode, the latter being the case of study of
turbance and limited bandwidth constraints. The cyber this article. Especially, large-scale penetration of solar PV gen-
networks are exposed to noisy disturbances and limited erators into an islanded microgrid system is becoming a popular
bandwidth constraints that terribly reduce the stability and
tradition rapidly [3], [4]. In this condition, the BESSs can act as
quality of the whole system. To eliminate the adverse ef-
fects of noisy disturbances and limited bandwidth con- the energy buffer in maintaining the power/energy balance and
straints, we propose a robust distributed control strat- overcoming the uncertainties of DERs and loads. This tradition
egy designed by using partial feedback linearization for arises several technical challenges among which the effective
the dynamical nonlinear model of a microgrid system. power flow control among DERs is the most significant, as
Moreover, a distributed event detection mechanism with
different types of generators are often geographically dispersed
noise-dependent threshold is adopted to update the control
signals with the consideration of unnecessary data com- and have different and limited power generation capabilities
munication reduction. Through adopting stochastic stabil- in providing active/reactive power support [5]–[7]. Then, ap-
ity theory and Lyapunov function, the stability and con- propriate power flow control strategies should be developed to
vergence analysis of the proposed dynamic distributed adjust the power injections of generators to satisfy the capacity
event-detection conditions considering noise interferences
constraint, building a desired distribution of the power outputs
is derived. As a result, the suggested method decreases
the sensitivity of the system to failures and increases its to meet the load demand under the power balance constraint and
reliability. Finally, a modified IEEE 34-bus test system in power generation constraint in the microgrid [8]–[10].
MATLAB/Simulink is utilized to verify the effectiveness of Since DERs are located in a heterogeneous and distributed
the proposed controller design scheme. fashion through low-voltage feeders, distributed control strate-
Index Terms—AC microgrid, cooperative control, dis- gies are required to improve the stability, scalability, and security
tributed event detection, dynamical nonlinear model, of the system [11], [12]. Distributed control strategy has become
mean-square exponential convergence, proportional power popular nowadays in controlling PV and BESSs to make the
sharing. power in ac microgrids easily dispatched and controlled, due
to its robustness to individual agent errors, scalability with re-
I. INTRODUCTION spect to increased number of agents, and reduced computational
load [13]–[15]. Actually, the communication network, working
O ACCOMMODATE the increasing penetration of re-
T newable energy sources, microgrids have been proposed
as interaction medium, plays an increasingly critical role in
distributed coordinate control for massive DERs.
Manuscript received October 21, 2019; revised December 8, 2019;
accepted December 27, 2019. Date of publication January 27, 2020; A. Related Research
date of current version October 23, 2020. This work was supported in
part by the National Natural Science Foundation of China under Grant In the majority of these researches, communication-scheduled
61773158, in part by the Fundamental Research Funds for the Central control protocols are often employed, that is, the distributed con-
Universities under Grant 2042019kf0186, in part by the Natural Science
Foundation of Hunan Province under Grant 2018JJ2051, and in part by trol signals are updated in a periodic sampling manner. However,
the Humboldt Research Council of Germany under Grant HB1807005. since the communication network of a distribution microgrid
Paper no. TII-19-4705. (Corresponding author: Xiaoqing Lu.) usually has limited bandwidth [16], [17], these communication-
J. Lai is with the E.ON Energy Research Center, RWTH Aachen
University, 52074 Aachen, Germany (e-mail: [email protected]). dependant control strategies [18], [19] have been proposed based
X. Lu is with the School of Electrical Engineering and Automation, on continuous communication among the distributed DERs,
Wuhan University, Wuhan 430072, China (e-mail: [email protected]). resulting in heavy burden on the communication channels. Thus,
Color versions of one or more of the figures in this article are available
online at https://ieeexplore.ieee.org. in order to alleviate the continuous communication burden
Digital Object Identifier 10.1109/TII.2020.2969458 on the communication channels for the distributed DERs, the
1551-3203 © 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.

Authorized licensed use limited to: M S RAMAIAH INSTITUTE OF TECHNOLOGY. Downloaded on October 20,2023 at 05:31:55 UTC from IEEE Xplore. Restrictions apply.
 220 IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS, VOL. 17, NO. 1, JANUARY 2021

works [20] and [21] presented the distributed control schemes sharing control performance in mean square. Compared with
for microgird regulation under event-triggered communication previous studies, the main advantages and contributions of the
to restore the voltage and frequency and achieve power sharing. proposed designs are summarized as follows.
To regulate power flow and voltage restoration of a dc microgird, 1) The exact model of PVs and BESSs based on feedback
a distributed event-triggered communication scheme was pre- linearization is considered in this article, which makes
sented in [22] by employing inter-DER communications. In [23], our results different from the previous works considering
an event-dependent distributed optimal control was addressed a superfluous model of microgrids [9]–[11]. Accordingly,
to achieve the voltage restoration and optimal current sharing the proposed robust nonlinear controllers can overcome
of a dc microgird, which is beneficial for the requirements of the limitations of linear controllers proposed in [14]–[18]
microgirds’ plug-and-play operation. Additionally, the ubiqui- by maintaining the power balance of microgrids over a
tous communication noise disturbances are inevitable in real wide range of operating conditions. The proposed robust
network transmission for the limited bandwidth, which may nonlinear power sharing control strategy employs a dis-
endanger the stability of microgrids and degrade their power tributed cooperation framework, which merely requires,
sharing performance. Recently, according to continuous com- when implemented only on local DER controllers and
munication, mean-square cooperative control was applied for DERs, a partial and limited knowledge of the problem
voltage/frequency synchronization of microgrids to attenuate parameters and can perform only local measurements,
this communication noise disturbances [24]–[26]. In fact, the which make the designed controllers easy to implement
influences of limited bandwidth constraints and noise distur- and reliable, and moreover, able to meet the plug-and-play
bances generally occur at the same time in practical commu- requirement.
nication networks. Furthermore, limited bandwidth constraints 2) Compared with existing ideal and periodic communi-
and noise disturbances are coupled with each other, which cation over microgirds [2], [7]–[10], even if the sparse
allows the effects of both limited bandwidth constraints and communication network suffers from noise disturbances
noise disturbances to be taken into account in the design of and limited bandwidth constraints, the proposed fully
the control scheme. Unfortunately, few works have been done nonlinear distributed control scheme can still achieve
for distributed power sharing regulation of massive DERs in mean-square proportional power sharing challenge
an ac microgird considering communication noises and limited among massive DERs. The network noise disturbances
bandwidth constraints simultaneously. and bandwidth constraints are simultaneously considered
On the other hand, most of the control methods [10]–[14] for in the proposed robust event-triggered control system,
microgrid systems treat DERs as ideal voltage sources behind which makes our results more meaningful yet practical.
the inverters and some of them consider the dynamics of inverters 3) Different from the existing distributed algorithms for
only in order to design proportional integral (PI) and other power sharing presented in [8], [9], and [11], the mean-
linear controllers. However, in an islanded mode, the system square proportional active and reactive power sharing will
dynamics are dictated by DERs, and therefore, it is essential be achieved under the power balance constraint and power
to consider the dynamics of DERs for an efficient operation generation constraint. Then, by using the stochastic anal-
of microgrids. Moreover, the linear controllers are unable to ysis techniques combined with the Lyapunov function,
provide appropriate active and reactive power with changes sufficient conditions on the event-detection conditions
in atmospheric conditions and system configurations such as considering noise disturbances are established to guar-
external disturbances and variations in loads as well as other antee the stability and reliability of the whole system.
system parameters [27]. Furthermore, the event-detection communication mecha-
Fortunately, nonlinear controllers overcome the limitations of nism can reduce costly computation for local controllers
linear controllers by maintaining the power balance of micro- and communication resources significantly.
grids over a wide range of operating conditions. The feedback
linearization technique transforms a nonlinear system into a fully
or partially linear system and linear control techniques can be C. Organization and Notations
employed to stabilize the whole system [21]–[24]. The genera- The remaining of this article is organized as follows.
tors in microgrids are connected very close to each other and the Section II presents the microgird control problem and provides
performance of one generator affects the dynamic performances some preliminaries. Section III discusses the proposed nonlin-
of others. Therefore, it is essential to consider the interactions ear mean-square distributed control scheme under distributed
among different generators, which can be done by implementing event detection for realizing autonomous proportional active
the feedback linearizing controllers in a distributed manner. and reactive power sharing among all DERs; furthermore, the
principle and stability analysis for the whole system are studied.
In Section IV, illustrative simulations are conducted on a 34-bus
B. Key Contributions
islanded system to validate the effectiveness of the proposed
With mainly the aforementioned inspirations, in this article, control scheme. Section V concludes this article.
a nonlinear mean-square power sharing control scheme under Notation: Denote || · || and ⊗ as the Euclidean norm and the
distributed event detection will be proposed to save the lim- Kronecker product, respectively. If A is a vector or a matrix,
ited network resource while preserving the autonomous power then AT denotes its transpose. If A is a matrix, denote by A

Authorized licensed use limited to: M S RAMAIAH INSTITUTE OF TECHNOLOGY. Downloaded on October 20,2023 at 05:31:55 UTC from IEEE Xplore. Restrictions apply.
 LAI AND LU: NONLINEAR MEAN-SQUARE POWER SHARING CONTROL FOR AC MICROGRIDS 221

Fig. 2. General configuration and local control diagram of DERi -PV

based on the VSIs in an islanded microgird.

Fig. 1. Proposed nonlinear distributed cooperative control structure for

a microgrid system.
aii = 0 otherwise. The Laplacian matrix L = (ij )N ×N (i, j ∈

V) is defined as ij = −aij , i = j and ii = N j=1,j=i aij ,
T
the spectral norm of A. As denotes A+A . λ2 (A) represents which has all row sums equal to zero and satisfies L1N = 0
2
the second smallest eigenvalue for a symmetric matrix A. tr(A) with 1N = (1, . . . , 1)T ∈ RN . The digraph G contains a span-
indicates the trace of a square matrix. ning tree if the eigenvalues of its Laplacian matrix satisfies
0 = λ1 (L) < Re(λ2 (L)) ≤ · · · ≤ Re(λN (L)) with Re(λi (L))
II. PROBLEM FORMULATION AND PRELIMINARIES  of the ith 
representing the real part eigenvalue of L. A digraph
G is called balanced if N i=j aij =
N
j=i aji for i = 1, . . . , N .
The proposed nonlinear distributed cooperative control For a balanced digraph, 1N is a left eigenvector of its Laplacian
scheme for an islanded ac microgrid system is shown in Fig. 1, matrix, i.e., 1TN L = 0.
where the microgrid system is composed of three layers—
electrical (physical) layer, control layer, and cyber (communi- B. DER System Model in Physical Layer
cation network) layer. The electrical layer includes the power
distribution network and power electronics interfacing. Such The detailed dynamical model based on feedback lineariza-
a physical system can be equipped with a cyber network to tion of a microgrid integrated with DER-PV and BESS subsys-
exploit different control paradigms. The control layer includes tems will be discussed in this subsection. As shown in Fig. 2,
the distributed control modules along with the primary control an inverter-based DER-PV subsystem is represented to supply
loop, and can be coupled with the physical network via sensors power into the microgrid. Furthermore, vector control is applied
and controllers embedded in power electronics devices. The in order to invert the rotating abc frame to the d–q frame and
cyber layer facilitates data exchange among different power elec- calculate the active and reactive powers from the converter to the
tronics inverters to collectively achieve consensus on qualities of grid. The DERi -PV subsystem that supplies power into the grid
interest. The cyber communication layer can be actualized and consists of a PV array in which a number of cells are connected
interfaced with the control layer using different communication in series and parallel combinations, a voltage-source inverter
protocols. A brief review on graph theory for communication (VSI) to convert the dc output power into the ac power, and an
topology and preliminary requirements (i.e., DER system model output filter to reduce the harmonics integrated with connecting
based on feedback linearization and local power of DER system) lines. By assuming that the output voltage of VSI can follow its
for the proposed controller is presented in the following subsec- voltage reference fast enough (i.e., viI = viDC∗ ), the dynamical
tion. model of a DERi -PV system [27] can be obtained from Fig. 2 in
d–q frame rotating with the common reference frequency ω com
of the microgrid, and in turn, can be represented as
A. Communication Topology in Cyber Layer
As demonstrated in Fig. 1, each DER in the microgrid is dviPV 1 PV 1 od 1 oq
= ii − ii − i
controlled by a local controller. The controllers exchange in- dt Ci Ci Ci i
formation with each other in a distributed manner to coordinate diod Rf PV
viod
n oq d vi
the operations of DERs in the microgrid. Therefore, the commu-
i
= − fi iod
i + ω i ii + γ i −
dt Li Lfi Lfi
nication topology in cyber layer formed by the DERs can also be
modeled as a directed graph G(V, ε) with the nodes or vertices dioq Rf PV
q vi vioq
i
= − fi ioq − ω n od
i + γ − (1)
V = {v1 , v2 , . . . , vN } and edges ε. The nodes in the graph dt Li i i i i
Lfi Lfi
represent the DERs, and the edges denote the communication
channel between the DERs. The condition (i, j) ∈ ε holds if where Rif and Lfi are the resistance and inductance of the filter
there is a link allowing the information flow from DERj to of the connecting lines, Ci is the dc-link capacitance, viDC is
DERi . Ni = {j ∈ V|(i, j) ∈ ε} is a set of DERi ’s neighbors, the voltage across the dc-link capacitor, viod , vioq and iod oq
i , ii
where j ∈ Ni if (i, j) ∈ ε. A = (aij )N ×N (i, j ∈ V) represents are, respectively, the feeder bus voltage vi and output current ioi
o

the graph adjacency matrix, where aij > 0 if (i, j) ∈ ε, and from inverter in d–q frame, iPV
i is the output current of DERi -PV

Authorized licensed use limited to: M S RAMAIAH INSTITUTE OF TECHNOLOGY. Downloaded on October 20,2023 at 05:31:55 UTC from IEEE Xplore. Restrictions apply.
 222 IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS, VOL. 17, NO. 1, JANUARY 2021

Similarly, the DERi -BESS system model as described in (2)

can be written in the following form of a generalized nonlinear
system:
ẋj = f (xj ) + g(xj )uj , yj = h(xj ) (4)
where xj = [i1j iod
j ioq T
j ] denotes the state vector of the DERj
d q T
unit, uj = [γj γj ] denotes the control input vector, yj =
[iod
j ioq T
j ] denotes the output vector
⎡ ⎤
Fig. 3. General configuration and local control diagram of DERj -BESS − τ11
⎢ j
⎥
based on the VSIs in an islanded microgird. ⎢ Rjf od vjod ⎥
fj (xj ) = ⎢ − Lf ij + ωj ioq − ⎥
n
⎢ j Lj ⎥
f

⎣ Rf oq ⎦
j
vj
− Lfj ioq
j − ω j ij + − L f
n od
array, and γid and γiq are the binary input switching signals, j j

respectively. and
As shown in Fig. 3, the DERj -BESS system consists of a ⎡ iod ioq
⎤
j j
battery, a VSI, and an output filter integrated with connecting ⎢ τj1 τj1 ⎥
lines [28]. The resistors (Rj0 and Rj1 ), capacitor (Cj1 ), and internal ⎢ vjDC ⎥
voltage (Ej ) are the specific internal structural parameters of a gj (xj ) = ⎢
⎢ Lfj 0 ⎥⎥.
⎣ DC
⎦
battery. Moreover, the charge stored in the battery is the integral 0
vj
of only a part i1j of the total current iDC entering the battery. Lfj
j
DERj -BESS can be modeled in a way similar to that of a The abovementioned nonlinear system models based on feed-
grid-connected DERj -PV system as discussed previously. In back linearization will be employed in the simulation and ver-
d–q frame, the dynamical model of a DERj -BESS as shown ification of the proposed mean-square control scheme in this
in Fig. 3 can be described as article. The dynamic characteristics of each node depend entirely
on the type of DER that the node accesses. That is, if a solar
di1j 1  oq 
j + ij − ij
= 1 iod 1
PV unit is connected at the node l, then the nonlinear system
dt τj models of node l are described as (3), and similarly, if a BESS
diod Rjf v DC vjod is connected at the node m, then the nonlinear system models
j n oq d j
= − f iod + ω i + γ − of node m are described as (4).
dt Lj j j j j
Lfj Lfj
dioq
j Rjf oq DC
q vj vjoq C. Local Power Control of DERs in Control Layer
= − f ij − ω j ij + γ j f − f
n od
(2)
dt Lj Lj Lj The measured active power and reactive powers of the DERi
unit can be derived as
where vjod , vjoq and iod oq
j , ij are, respectively, the feeder bus oq oq oq od od oq
voltage vj and output current ioj from the inverter in d–q frame; i + v i ii , Q i = v i ii − v i ii
Pi = viod iod
o (5)
γjd and γjq are the switching functions in d–q frame, respectively. where Pi and Qi are the measured active and reactive powers,
Employing the feedback linearization technique, the DERi -PV respectively.
system model as described in (1) can be written in the following Under d–q rotating frame, the terminal voltage vio of the DERi
form of a generalized nonlinear system: units satisfies viod = vio and vioq = 0. Accordingly, the local
ẋi = f (xi ) + g(xi )ui , yi = h(xi ) (3) power control (5) of the DERi in d–q frame can be simplified as
od oq
where xi = [viDC iod ioq T
ui = [γid γiq ]T , and yi = i , Qi = −vi ii .
Pi = viod iod (6)
i i ] ,
od oq T
[ii ii ] , respectively, denote the state vector, the control From (6), we can see that the active and reactive powers Pi and
input vector, and the output vector oq
Qi can be controlled by the current iodi and ii , respectively.
⎡ Rif PV
⎤ As shown in Fig. 2, the dynamics of the proportional integral
i i
⎢ Ci
⎥ (PI) current controller with the cancellation of the d–q cross-
⎢ Rif od v od ⎥
fi (xi ) = ⎢ − Lf ii + ωin ioq − Li f ⎥ coupling can be described as
⎣ i
i i ⎦  
Rf viod viId = viod − ωicom Lfi ioq iiref − iod
iP od
− Lfi ioq
i − ω n od
i
i i − Lf
i + ki i
i i
 
+ kiiI iref − ii
iod
and od
dt
⎡ ⎤
− C1 iod
i − C1 ioq
i
⎢ ⎥ viIq = vioq + ωicom Lfi iod iP oq oq
i + ki (iiref − ii )
i i
viPV
gi (xi ) = ⎢
⎣
0 Lfi
⎥.
⎦
viPV (ioq oq
Lfi
0 + kiiI iref − ii ) dt (7)

Authorized licensed use limited to: M S RAMAIAH INSTITUTE OF TECHNOLOGY. Downloaded on October 20,2023 at 05:31:55 UTC from IEEE Xplore. Restrictions apply.
 LAI AND LU: NONLINEAR MEAN-SQUARE POWER SHARING CONTROL FOR AC MICROGRIDS 223

where kiiP and kiiI are the proportional gain and integral gain of reactive power values. Accordingly, the interconnected
output current ioi , respectively. DERs can achieve mean-square proportional power sharing,
Pj
The reference error signals of the active and reactive pow- which means maintaining limt→+∞ E P Pmax i
− P max 2 =
i j
ers are obtained from the power flow information based on Qj
0, limt→+∞ E QQmax
i
− Qmax 2 =0 for all i = j ∈
the autonomy achieved by different neighboring subsystems, i j

which will be designed in Section III in detail. Accordingly, the {1, 2, . . . , N }.

proposed distributed controller design translates the active and The main objective of power sharing is to maintain a balance
reactive powers into the current reference signals (i.e., iod between the generation and the demand within the systems under
iref and
ioq ) in the reference d–q frame as follows: the power balance constraint and power generation constraint
iref
N
χP
ave Pi
max
χQ
ave Qi
max
iod
iref = od
, ioq
iref = − od
(8) Pi =PD Pimin ≤ Pi ≤ Pimax (10)
vi vi
i=1

where χP Q
ave and χave are autonomous proportional active and where PD denotes the total power demand. Assume that all DERs
reactive power sharing ratios at the steady state. are operating within their generation constraints. Considering a
converter with the maximum capacity of Simax and the active
III. NONLINEAR MEAN-SQUARE POWER SHARING CONTROL power of Pimax , the maximum available reactive power Qmax i
FOR AC MICROGRIDS UNDER DISTRIBUTED of converter can be expressed as
EVENT DETECTION
To ensure the autonomous proportional active and reactive Qmax
i = (Simax )2 − (Pimax )2 . (11)
power sharing, a nonlinear mean-square cooperative control
scheme under distributed event detection will be developed, Based on the abovementioned approach, a DER unit which
which is fully distributed and only triggered at their own produces lower amount of active power is expected to produce
event time. Accordingly, it effectively reduces the frequency higher amount of reactive power. It is worth noting that the
of controller updates compared with continuous-time feedback maximum capacity limit, based on (11), is applied to the system
control, and moreover, is also robust to noise disturbances. in order to avoid overloading the DER units.
This section presents the synthesis of the autonomous power For the above-mentioned purpose, recall the dynamic (3) and
sharing inputs for the DERs by the so-called consensus approach (4), it is clear that Pi , Qi and Pj , Qj can be, respectively, tuned by
of multiagent systems [16], [17], i.e., DER devices exchange controlling the direct and quadrature terms of the output currents,
oq oq oq
information with others to achieve consensus by themselves iod
i , ii and ij , ij . By differentiating (9), the proportional
and without external commands. Eventually, it will benefit the power sharing problem can be treated as the following consensus
utilization of the DERs and improve the power supply reliability problem:
and flexibility. Ṗi Q̇i
= uP
i , = uQ
i (12)
Pimax Qmax
i
A. Proportional Power Sharing of All DERs
Q
Since different DERs have different capacities in a microgrid, for i = 1, . . . , N , where uP
i and ui are two nonlinear controllers
it is desirable to share the overall load powers according to designed in the following subsection.
their respective capacities. To this end, in this article, all DERs
in a microgrid are controlled to operate with the same active B. Nonlinear Mean-Square Autonomous Proportional
and reactive power output ratios by themselves and without Power Sharing via Distributed Event Detection
external commands. As a result, the utilization level of different
It is found that the conventional ideal and periodic communi-
DERs can be identical in steady states. Namely, all DERs are
cation methods are utilized for the proposed proportional power
autonomous with the same active and reactive power output Q
sharing controllers uP i and ui to achieve the autonomous power
ratios as follows:
utilization ratio, which may bring significant communication
Pi Pj Qi Qj burden and may not be robust against noise disturbances. In
max = max = χP ave , max = max = χQ ave (9)
Pi Pj Qi Qj order to achieve the autonomous proportional active and reactive
power sharing, each DER only needs to communicate with its
where P Pmax
i
and QQmax
i
denote the output ratios of active and neighbors through a sparse communication network with noise
i i
reactive powers, which implies the utilization percentages of disturbances and limited bandwidth constraints. To this end,
the DERi unit; Pimax and Qmax
i denote the maximum capacities we first assume that the sequence of event times for DERi is
of Pi and Qi , respectively. χP Q
ave and χave are autonomous 0 = ti0 < ti1 < ti2 < · · · < tiθ < · · · (θ = 0, 1, 2, . . .), moreover
proportional active and reactive power sharing ratios at the the DERi broadcasts its state only at its own event times. Thus,
steady state, which will automatically converge to the average there is no necessity for DERi to measure its neighbor’s state
N Pi (0) N Qi (0)
power ratios (i.e., N1 i = 1 Pimax and N
1
i = 1 Qmax ) by information in real time. Accordingly, the proposed nonlinear
i
Q
the designed robust distributed control in the following, mean-square controllers uP i and ui will be developed by em-
Pi (0) and Qi (0) are, respectively, the initial active and ploying the last measurement information received from each

Authorized licensed use limited to: M S RAMAIAH INSTITUTE OF TECHNOLOGY. Downloaded on October 20,2023 at 05:31:55 UTC from IEEE Xplore. Restrictions apply.
 224 IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS, VOL. 17, NO. 1, JANUARY 2021

neighbor j ∈ Ni as follows: Thus, from (9) to (13), we can have the following error system:
 
Pj (tjk ) Pi (tik ) de(t) = [−c(t)L(e(t) + δ(t))]dt + c(t)(I − Δ)Σdη(t). (15)
P
ui (t) = c(t) aij − max + σij ηij
Pjmax Pi Consider a Lyapunov function as V (t) = 21 εT (t)ε(t) with
j∈Ni
  ε(t) = Le(t) for t ∈ [tk , tk + 1 ), then, based on the Itô formula,
Q Qj (tjk ) Qi (tik ) the stochastic derivative of V along trajectories of system (15)
ui (t) = c(t) aij − max + σij ηij (13)
Qmax
j Qi can be expressed as
j∈Ni
dV (t) = εT (t)L[−c(t)(ε(t) + Lδ(t))]dt
such that
 2 + εT (t)L[c(t)(I − Δ)Σ]dη(t) + c2 (t)Ξdt (16)
 P (t) Pi (t) 
 j 
lim E  − max  = 0 where Ξ = tr(L(I − Δ)Σ2 (I − Δ)LT )/2.
t→+∞  P max Pi 
j
Furthermore, in order to implement the developed nonlinear
 2
 Q (t) Q (t)  distributed cooperative controllers based on the event-triggered
 j i 
lim E  max − max  = 0 (14) mechanism, an event detector is employed for each slave unit to
t→+∞  Q Q 
j i determine when the sampled local information should be utilized
to update the control actions of itself and its neighbors. The
for all i = 1, . . . , N , where t ∈ [tik , tik+1 ), tik = max{ti |ti ∈
following event detection condition can guarantee that
{tiθ }, ti ≤ t} and tjk = max{tj |tj ∈ {tjθ }, tj ≤ tik } are the   
 
transmitted messages of DERi and its neighbors at their  Pj (t) Pi (t) 
last event times, respectively. c(·) : [0, ∞) → (0, ∞) is a δi (t) ≤ σij 
 a −  (17)
Pimax 
ij
j∈Ni Pjmax 
consensus-gain function [26], which is piecewise continuous
and nonincreasing.
∞ The consensus-gain function c(·) also sat-
where 0 < κ < 1, 0 < σij ≤ κλ2 (Ls )/L2 .
isfies
∞ 2 0 c(s)ds = ∞, called the convergence condition, and
Suppose that the balanced digraph G always contains a
c (s)ds < ∞, called the robustness condition. In fact, the
0 spanning directed tree, the matrix Ls will be the Laplacian
consensus-gain function is employed to attenuate the measure-
matrix for the mirror graph of G, which is undirected and
ment noises, and there are many functions that meet these
connected. Moreover, for a balanced digraph, 1T L = 0 holds,
properties, for example, c(t) = t+1 1
. {ηij , i, j = 1, 2, . . . , N }
then 1T ε(t) = 1T Le(t) = 0 will be satisfied. Then, we can get
are standard white noises and σij ≥ 0 is the noise intensity.
εT (t)L[−c(t)(ε(t) + Lδ(t))]
C. Stability Analysis
= −c(t)εT (t)Ls ε(t) − c(t)εT (t)LLδ(t)
In the following, the graph theory, stochastic theory, and
Lyapunov technique are employed to derive some stability cri- ≤ −c(t)λ2 (Ls )ε2 + c(t) ε L Lδ
teria to guarantee feasibility of the microgrid model. Define the ≤ −c(t)λ2 (Ls )ε2 + c(t) ε κλ2 (Ls ) Le
following two types of local state measurement errors δi (t) =
ρi (tik ) − ρi (t) and δj (t) = ρj (tjk ) − ρj (t), for DERi , j ∈ Ni , = −c(t)(1 − κ)λ2 (Ls )ε2 . (18)
where ρi (t) = PPimax (t)
. The local state measurement error repre-
i Accordingly, (16) satisfies
sents the output frequency deviation that the present time states  
deviate from the last sample time states, and when the local state dV (t) ≤ −2c(t)(1 − κ)λ2 (Ls )V (t) + c2 (t)Ξ dt
measurement error of the DERi reaches a time-varying threshold
prescribed in the following, the event is triggered, and then, + c(t)εT (t)L(I − Δ)Σdη(t) (19)
the DERi transmits the measured state and begins to update its for any t ∈ [tk , tk + 1 ), k = 0, 1, . . ..
controllers. Accordingly, let δ(t) = (δ1 (t), . . . , δN (t))T ∈ RN . Furthermore, for any t ≥ 0 and h > 0, from [25], (19) yields
Denote Σi = diag{σi1 , . . . , σiN } for i = 1, . . . , N . Let
a(i, ·) be the ith row of the adjacency matrix A = (aij )N ×N E[V (t + h)] − E[V (t)]
lim sup
of the digraph G and matrix h→0+ h
⎛ ⎞ ≤ −2c(t)(1 − κ)λ2 (Ls )E[V (t)] + c2 (t)Ξ. (20)
a(1, ·)Σ1
⎜ .. ⎟ N ×N 2 Consequently, according to the comparison principle [26], we
Σ=⎝ . ⎠∈R .
obtain
a(N, ·)ΣN
E[V (t)] ≤ ΞΩ1 (t) + Ω2 (t) (21)
Then, the global neighborhood tracking error for  τ0  t
the output activepower of DERi can be defined as where Ω1 (t) = 0 exp{−2(1 − κ)λ2 (Ls ) τ c(r)d(r)}c2 (τ )
N τ t
ei (t) = ρi (t) − N1 i = 1 ρi (t), for i = 1, . . . , N , and e(t) = dτ and Ω2 (t) = V (0) 0 0 exp{−2(1 − κ)λ2 (Ls ) 0 c(τ )d(τ )}.
1N T
(e1 (t), . . . , eN (t)) ∈ RN , the matrix Δ = 1NN
T
, then we According to consensus-gain function defined in [26],
have e(t) = ρ(t) − Δρ(t) with ρ(t) = (ρ1 (t), . . . , ρN (t))T . limt→∞ Ω2 (t) = 0 and λ2 (Ls ) > 0. Moreover, for given ζ > 0,

Authorized licensed use limited to: M S RAMAIAH INSTITUTE OF TECHNOLOGY. Downloaded on October 20,2023 at 05:31:55 UTC from IEEE Xplore. Restrictions apply.
 LAI AND LU: NONLINEAR MEAN-SQUARE POWER SHARING CONTROL FOR AC MICROGRIDS 225

∞
there exists τ0 > 0 satisfying τ0 c(τ )dτ < ζ. Thus, for arbitrary
t > τ0 , we have
τ0 t
Ω1 (t) = exp{−2(1 − κ)λ2 (Ls ) c(r)d(r)}c2 (τ )dτ
0 τ
∞ t
+ exp{−2(1 − κ)λ2 (Ls ) c(r)d(r)}c2 (τ )dτ
τ0 τ
τ0 t
≤Θ c2 (τ )dτ + c2 (τ )dτ
0 τ0
∞ ∞
≤Θ c2 (τ )dτ + c2 (τ )dτ
0 τ0
∞
≤Θ c2 (τ )dτ + ζ (22)
0
t
with Θ = exp{−2(1 − κ)λ2 (Ls ) τ0 c(r)d(r)}.
Accordingly, limt→∞ ΞΩ1 (t) = 0 can be obtained, due to
the arbitrariness of ζ. Then, limt→∞ E[V (t)] = 0 holds, which
means that limt→∞ E[e(t)2 ] = 0. In this situation, we get
N Pi (0) 2
limt→∞ E[ PPimax
(t)
− N1 i = 1 P max  ] = 0, i.e., for all i ∈
i i

{1, . . . , N }, the active power output PPimax

(t)
converge to χP
ave =
1
N Pi (0) i

N i=1 Pimax asymptotically in mean square by Lyapunov

stability theorem. Accordingly, we can achieve autonomous Fig. 4. Proposed autonomous power sharing control structure imple-
mented for the ith DER in an islanded microgird via a sparse cyber
proportional active power sharing by supplying χP ave for local network.
power control in (8). Furthermore, we can achieve autonomous
proportional active power sharing through replacing P slightly
with Q. Eventually, the conclusion of the proposed nonlinear
mean-square distributed power sharing controllers can be de- Remark 1: Different from the related works based on the
scribed in Theorem 1. existence of a virtual leader [13]–[17], [19]–[26], this article
Theorem 1: Suppose that the balanced digraph G always proposes a possible solution to achieve self-consistent pro-
contains a spanning directed tree. If the event detection condition portional active and reactive power sharing among all DERs
(17) is satisfied, then the stochastic distributed controllers (13) through neighboring information exchange without any external
combined with event-triggered communication with noise inter- commands. Actually, since there exists the uncertainty of the
ferences can achieve the output ratio autonomous of active and communication network, it is necessary to consider the case
reactive power sharing, P Pmax i Pj
= P max and QQmax
i Qj
= Qmax (for when the leader fails. In this situation, the previously proposed
i j i j methods [19]–[26] may degrade or even destroy the system
all i = j), asymptotically. Finally, the autonomous proportional
stability. However, this article proposes a kind of self-organizing
active and reactive power sharing among all DERs in a microgrid
idea-based possible solution to achieve self-consistent propor-
can be achieved.
tional active and reactive power sharing among all DERs. Each
Implementation of the proposed nonlinear mean-square au-
DER can be recognized as an agent, it can regulate its own
tonomous control scheme under distributed event detection is
states according to its neighbors’ storage devices, and exchange
illustrated in Fig. 4. At each sampling instant, through the
information with others to achieve consensus by themselves and
communication network in cyber layer, each DER broadcasts its
without external commands.
state information to the neighbors and also receives state infor-
mation from its neighbors with communication noises σij ηij for
computing by (17), which is employed for event detection. If the
IV. CONTROLLER PERFORMANCE VALIDATION
condition in (17) is satisfied, no further action is required for this
DER; otherwise, this DERi unit will update its control actions In order to demonstrate the performance of the proposed
Q
uPi and ui and notify its neighbors to, respectively, update control scheme, the modified IEEE 34-bus test microgrid [29]
their control actions P Pmax i
and QQmax
i
by using its current state as shown in Fig. 5, is simulated in MATLAB/Simulink. Other
i i
Pi (tik ) Qi (tik ) used parameters of the test microgrid system are shown in
information (i.e., Pimax and Qmax ). Accordingly, the control Table I. A local communication network providing its required
i
Q
actions uP od
i and ui are to generate the current reference iiref and information flows will be employed for the proposed nonlinear
oq
iiref by (8), for the proportional integral (PI) current control loop. distributed power sharing controllers. This article studies the
Finally, the current errors are calculated and used to regulate the stability and performance of a microgrid under various sce-
outputs of the inverter by PWM mode. narios. The associated adjacency matrix of G of the microgird

Authorized licensed use limited to: M S RAMAIAH INSTITUTE OF TECHNOLOGY. Downloaded on October 20,2023 at 05:31:55 UTC from IEEE Xplore. Restrictions apply.
 226 IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS, VOL. 17, NO. 1, JANUARY 2021

Fig. 5. Single line diagram of the DERs and local loads in a IEEE
34-bus microgrid with the communication network topology G.

TABLE I
PARAMETERS FOR THE TEST MICROGRID SYSTEM

system is
⎛ ⎞
0 1 0 0 1
⎜ ⎟
⎜1 0 1 0 0⎟
⎜ ⎟
A=⎜
⎜0 1 0 1 0⎟⎟.
⎜ ⎟
⎝0 0 1 0 1⎠
1 0 0 1 0
The additive noise is added with σij = 0.4 in the whole
communication process, take κ = 0.5, which satisfies 0 < σij ≤
κλ2 (Ls ) = 0.528.

A. Performance Assessment With Various Load

Demand
In this case, the performance of the proposed controllers
in different communication digraphs is designed in the situa-
tion of load changes. Communication topology for this case
is illustrated in Fig. 6. At t = 0 s, the system starts with the Fig. 6. Performance of the proposed mean-square autonomous power
proposed controllers (13). Meanwhile, the communication links sharing control in case of load change with additive noise σij = 0.4:
between DER1 and DER5 , DER2 , and DER3 in digraph G is (a) active power ratio P P i
max ; (b) active power output; (c) reactive power
i
Qi
disconnected. In this situation, we know that DER1 can only get ratio Qmax ; and (d) reactive power output.
i
the information from DER2 . Similarly, there exists a spanning
tree among DER3 , DER4 , and DER5 , then they can achieve the
information exchanges among them. Accordingly, it is shown in whole system being able to achieve the information exchanges.
Fig. 6 that the designed controllers can achieve the autonomous Accordingly, all DERs can achieve autonomous proportional ac-
proportional active and reactive power sharing among DER1 and tive and reactive power sharing through the cooperation among
DER2 , DER3 , DER4 , and DER5 , respectively. Subsequently, at them.
t = 5 s, the communication links between DER1 and DER5 , Then, at t = 9 s, the local Load3 decreased from full load
DER2 and DER3 in digraph G is reconnected, which leads to the to 0. In turn, Load3 increased from 0 to full load at t = 12

Authorized licensed use limited to: M S RAMAIAH INSTITUTE OF TECHNOLOGY. Downloaded on October 20,2023 at 05:31:55 UTC from IEEE Xplore. Restrictions apply.
 LAI AND LU: NONLINEAR MEAN-SQUARE POWER SHARING CONTROL FOR AC MICROGRIDS 227

Fig. 7. Proposed mean-square autonomous power controllers acti-

vated at the time instant [0 s, 15 s] with event-triggered broadcast
periods and periodic broadcast periods.

s. As seen, the power utilization ratios of each DER vary

with the change of local loads within the microgrid system,
however, the autonomous power ratio among DERs is always
maintained, as shown in Fig. 6(a) and (b). The interaction
during autonomous and load change is smooth since DER can
respond quickly; therefore, Fig. 6 illustrates that the proposed
distributed power sharing controllers have the ability to make
DERs achieve the steady state rapidly. The proposed design
can achieve the autonomous proportional power sharing among
DERs when the communication structure among them varies but
is still connected. This is a necessary condition in the proposed
Theorem 1.

B. Performance Comparative Analysis

To demonstrate that the proposed mean-square distributed
control strategy under an event-detection mechanism can reduce
the communication amount between DERs, the comparative
analysis on event time instant and periodic time instant for
DERs under the load change case is given in Fig. 7. Under
the event-detection mechanisms (13), the event instants and
data release intervals of DERs are illustrated in Fig. 7, showing
clearly that the number of DER control updates is sporadic rather
than every time instant. During the 15-s simulation period with
the proposed controllers activated, the communication numbers
of DER1 –DER5 generated by these event-triggered communica-
tion mechanisms are 3825. It implies that only 51% (3825/7500)
sampled-data exchange is needed.
Eventually, the performance of the proposed mean-square
distributed control scheme is compared with that of a previously
proposed distributed control method in [15]. We perform the
simulation scenario of Section IV-A by exploiting the distributed
controller proposed in [15], and present the results in Fig. 8. As
seen, using the previously proposed controllers in [15], when
the additive noise is added with σij = 0.2 at t = 5 s, after that Fig. 8. Performance evaluation of distributed controllers using the
the power sharing output waveforms become unstable. In con- previously proposed distributed control method: (a) active power ratio
Pi Qi
trast, as discussed in Section IV-A, the proposed mean-square P max ; (b) active power output; (c) reactive power ratio Qmax ; and (d)
i i
distributed controller has desirable performance. Furthermore, reactive power output.
when the additive noise is changed to σij = 0.4 at 9 s, as seen
in Fig. 8, the unstability of power sharing output waveforms in-
C. Plug-and-Play Capability
creases even further. Eventually, unlike the previously proposed
control method in [15], one of the very expected advantages of The plug-and-play capability is important for a microgrid due
the proposed control scheme lies in the consensus-gain
∞ func- to the intermittent nature of DERs. In other words, when some
tion
∞ 2 c(t) adopted in (13), which satisfies 0 c(s)ds = ∞ and DERs are plugged into or out of the microgrid, the proposed
0 c (s)ds < ∞ to attenuate the measurement noises. controllers are still effective to make the whole system operate

Authorized licensed use limited to: M S RAMAIAH INSTITUTE OF TECHNOLOGY. Downloaded on October 20,2023 at 05:31:55 UTC from IEEE Xplore. Restrictions apply.
 228 IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS, VOL. 17, NO. 1, JANUARY 2021

property and maintain transient stability of microgrid, then the

distributed power sharing control can achieve the autonomous
proportional power sharing among the remaining DERs. Since
DER5 is plugged out at t = 10 s, the whole system will lose
active power 30 kW and reactive power 20 kVAr, which leads to
an increase in the output power ratios of the remaining DERs to
compensate for these power deficiencies as shown in Fig. 9(a)
and (c). At the same time, the corresponding output powers
also increase proportionally as shown in Fig. 9(b) and (d). It
should be noted that when a DER is plugged out, it also implies
loss of all communication links with its neighbors in the cyber
layer. Accordingly, when DER5 is plugged out, it automatically
renders the links DER1 –DER5 and DER4 –DER5 . However,
the remaining links still contain a spanning tree. Subsequently,
DER5 with communication link establishment is plugged back
at t = 15 s. As seen, the controllers have properly updated the
power sharing after DER5 is plugged back at the steady state.
It is worth noting that although DER5 is unplugged instantly,
the power measurements exponentially decay to zero due to the
existence low-pass filters.

V. CONCLUSION
A nonlinear mean-square proportional power sharing control
strategy implemented DERs system was investigated in this arti-
cle, where the distributed communication network suffers from
additive noise and limited bandwidth constraints. The proposed
controllers achieved all DERs mean-square proportional power
sharing based on their status and capacities by autonomously ad-
justing itself. To further reduce the communication requirement,
an event-detection mechanism that can be evaluated locally was
introduced where each local controller only communicates to its
neighbors when the detecting condition is satisfied. Moreover,
the proposed controllers are also robust to noise disturbances.
The inequality technique synthesis and simulation with the
modified test microgrid model demonstrated the effectiveness
of the proposed control algorithm.

REFERENCES
[1] X. Lu, X. Yu, J. Lai, J. M. Guerrero, and Y. Wang, “A novel distributed
secondary coordination control approach for islanded microgrids,” IEEE
Trans. Smart Grid, vol. 9, no. 4, pp. 2726–2740, Jul. 2018.
[2] F. Wang, K. Li, C. Liu, Z. Mi, M. Shafie-Khah, and J. P. S. Catalão, “Syn-
chronous pattern matching principle based residential demand response
baseline estimation: Mechanism analysis and approach description,” IEEE
Trans. Smart Grid, vol. 9, no. 6, pp. 6972–6985, Nov. 2018.
Fig. 9. Performance of the proposed mean-square autonomous power [3] J. Lai, X. Lu, X. Yu, and A. Monti, “Cluster-oriented distributed coop-
sharing control in case of plug-and-paly of DERs with additive noise erative control for multiple ac microgrids,” IEEE Trans. Ind. Informat.,
σij = 0.4: (a) active power ratio P P i
max ; (b) active power output; (c)
vol. 15, no. 11, pp. 5906–5918, Nov. 2019.
i [4] Q. Yang, J. A. Barria, and T. C. Green, “Communication infrastructures
Qi
reactive power ratio Qmax ; and (d) reactive power output. for distributed control of power distribution networks,” IEEE Trans. Ind.
i
Informat., vol. 7, no. 2, pp. 316–327, Feb. 2011.
[5] K. Shu et al., “Real-time subsidy based robust scheduling of the integrated
in a normal mode. To address this concern, it is assumed that power and gas system,” Appl. Energy, vol. 236, 1158–1167, 2019.
[6] Y. A. R. I. Mohamed, H. Zeineldin, M. Salama, and R. Seethapathy, “Seam-
DER5 is physically plugged out and then plugged back to the less formation and robust control of distributed generation microgrids via
microgrid system at t = 10 s and at t = 15 s, respectively. direct voltage control and optimized dynamic power sharing,” IEEE Trans.
Initially, the operating conditions of the whole system are the Power Electron., vol. 27, no. 3, pp. 1283–1294, Mar. 2012.
[7] X. Lu, J. Lai, X. Yu, Y. Wang, and J. M. Guerrero, “Distributed coordination
same as that described in part A of Section IV. It is observed in of islanded microgrid clusters using a two-layer intermittent communica-
Fig. 9 that, at t = 10 s, the proposed controllers (13) can render tion network,” IEEE Trans. Ind. Informat., vol. 14, no. 9, pp. 3956–3969,
the dynamic changes caused by the effect of plug-and-play Sep. 2018.

Authorized licensed use limited to: M S RAMAIAH INSTITUTE OF TECHNOLOGY. Downloaded on October 20,2023 at 05:31:55 UTC from IEEE Xplore. Restrictions apply.
 LAI AND LU: NONLINEAR MEAN-SQUARE POWER SHARING CONTROL FOR AC MICROGRIDS 229

[8] A. Milczarek, M. Malinowski, and J. M. Guerrero, “Reactive power man- [25] J. Lai, X. Lu, A. Monti, and G. P. Liu, “Stochastic distributed pin-
agement in islanded microgrid—Proportional power sharing in hierarchi- ning control for co-multi-inverter networks with a virtual leader,” IEEE
cal droop control,” IEEE Trans. Smart Grid, vol. 6, no. 4, pp. 1631–1638, Trans. Circuits Syst. II, Express Briefs, to be published, doi: 10.1109/TC-
Jul. 2015. SII.2019.2950764.
[9] S. Dasgupta, S. N. Mohan, S. K. Sahoo, and S. K. Panda, “Lyapunov [26] J. Lai, X. Lu, and X. Yu, “Stochastic distributed frequency and load sharing
function-based current controller to control active and reactive power flow control for microgrids with communication delay,” IEEE Syst. J., vol. 13,
from a renewable energy source to a generalized three-phase microgrid no. 4, pp. 4269–4280, Dec. 2019.
system,” IEEE Trans. Ind. Electron., vol. 62, no. 2, pp. 799–813, Feb. 2013. [27] M. A. Mahmud, H. R. Pota, and M. Hossain, “Dynamic stability of three-
[10] P. H. Huang, P. C. Liu, W. Xiao, and M. S. El Moursi, “A novel droopbased phase grid-connected photovoltaic system using zero dynamic design
average voltage sharing control strategy for dc microgrids,” IEEE Trans. approach,” IEEE J. Photovolt., vol. 2, no. 4, pp. 564–571, Oct. 2012.
Smart Grid, vol. 6, no. 3, pp. 1096–1106, May 2015. [28] M. A. Mahmud, H. R. Pota, M. Aldeen, and M. Hossain, “Partial feed-
[11] A. Kahrobaeian and Y. A.-R. I. Mohamed, “Networked-based hybrid back linearizing excitation controller for multimachine power systems
distributed power sharing and control for islanded microgrid systems,” to improve transient stability,” IEEE Trans. Power Syst., vol. 29, no. 2,
IEEE Trans. Power Electron., vol. 30, no. 2, pp. 603–617, Feb. 2015. pp. 561–571, Mar. 2014.
[12] J. Fang, Q. Zeng, X. Ai, Z. Chen, and J. Wen, “Dynamic optimal energy [29] J. Lai, X. Lu, F. Wang, P. Dehghanian, and R. Tang, “Broadcast gossip
flow in the integrated natural gas and electrical power systems,” IEEE algorithms for distributed peer-to-peer control in ac microgrids,” IEEE
Trans. Sustain. Energy, vol. 9, no. 1, pp. 188–198, Jan. 2018. Trans. Ind. Appl., vol. 55, no. 3, pp. 2241–2251, May/Jun. 2019.
[13] X. Lu, X. Yu, J. Lai, and Y. Wang, “Finite-time control for robust tracking
consensus in mass with an uncertain leader,” IEEE Trans. Cybern., vol. 47,
no. 5, pp. 1210–1223, May 2017.
[14] J. W. Simpson-Porco, F. Dorfler, and F. Bullo, “Synchronization and power
sharing for droop-controlled inverters in islanded microgrids,” Automatica,
vol. 49, no. 9, pp. 2603–2611, Aug. 2013. Jingang Lai (Senior Member, IEEE) received
[15] J. Lai, H. Zhou, X. Lu, X. Yu, and W. Hu, “Droop-based distributed the Ph.D. degree in control science and engi-
cooperative control for microgrids with time-varying delays,” IEEE Trans. neering from Wuhan University, Wuhan, China,
Smart Grid, vol. 7, no. 4, pp. 1775–1789, Jul. 2016. in 2016.
[16] J. Lai, X. Lu, X. Yu, A. Monti, and H. Zhou, “Distributed voltage regulation In 2015, he was a Joint Ph.D. Student with the
for cyber-physical microgrids with coupling delays and slow switching School of Electrical and Computer Engineer-
topologies,” IEEE Trans. Syst., Man, Cybern., Syst., vol. 50, no. 1, pp. 100– ing, RMIT University, Melbourne, VIC, Australia.
110, Jan. 2020. He is currently a Humboldt Research Fellow
[17] X. Lu, X. Yu, J. Lai, J. M. Guerrero, and H. Zhou, “Distributed secondary with the E.ON Energy Research Center, RWTH
voltage and frequency control for islanded microgrids with uncertain com- Aachen University, Aachen, Germany. His re-
munication links,” IEEE Trans. Ind. Informat., vol. 13, no. 2, pp. 448–460, search interests include distributed intelligence
Apr. 2017. for ac/dc microgirds, distributed renewable energy system applications
[18] P. Prabhakaran, Y. Goyal, and V. Agarwal, “A novel communication-based of multiagent systems, and cyber–physical networked control systems.
average voltage regulation scheme for a droop controlled dc microgrid,” Dr. Lai was the recipient of the Finally List Paper Certificate for the
IEEE Trans. Smart Grid, vol. 10, no. 2, pp. 1250–1258, Mar. 2019. IEEE Conference on Industrial Electronics and Applications in 2017. He
[19] C. Yang et al., “Dynamic event-triggered robust secondary frequency is currently an Associate Editor for the IEEE TRANSACTIONS ON INDUSTRY
control for islanded ac microgrid,” Appl. Energy, vol. 242, pp. 821–836, APPLICATIONS and IET Generation, Transmission and Distribution.
2019.
[20] J. Lai, X. Lu, X. Yu, and A. Monti, “Stochastic distributed secondary
control for ac microgrids via event-triggered communication,” IEEE Trans.
Smart Grid, to be published, doi: 10.1109/TSG.2020.2966691.
[21] D. Pullaguram, S. Mishra, and N. Senroy, “Event-triggered communication
based distributed control scheme for dc microgrid,” IEEE Trans. Power Xiaoqing Lu (Senior Member, IEEE) received
Syst., vol. 33, no. 5, pp. 5583–5593, Sep. 2018. the Ph.D. degree in applied mathematics from
[22] W. Meng, X. Wang, and S. Liu, “Distributed load sharing of an inverter- Wuhan University, Wuhan, China, in 2012.
based microgrid with reduced communication,” IEEE Trans. Smart Grid, From 2015 to 2019, she was a Postdoc-
vol. 9, no. 2, pp. 1354–1364, Mar. 2018. toral Fellow with the School of Engineering,
[23] J. Lai, X. Lu, X. Yu, W. Yao, J. Wen, and S. Cheng, “Distributed multi-DER RMIT University, Melbourne, VIC, Australia. She
cooperative control for master-slave-organized microgrid networks with is currently a Professor with the School of
limited communication bandwidth,” IEEE Trans. Ind. Informat., vol. 15, Electrical Engineering and Automation, Wuhan
no. 6, pp. 3443–3456, Jun. 2019. University. Her research interests include dis-
[24] S. Shrivastava, B. Subudhi, and S. Das, “Noise-resilient voltage and tributed control, monitoring, and security in
frequency synchronisation of an autonomous microgrid,” IET Gener. cyber-physical systems; and cooperation and
Transmiss. Distrib., vol. 13, no. 2, pp. 189–200, 2019. optimization in smart microgrids and multi-robot systems.

Authorized licensed use limited to: M S RAMAIAH INSTITUTE OF TECHNOLOGY. Downloaded on October 20,2023 at 05:31:55 UTC from IEEE Xplore. Restrictions apply.