Two-Stage Optimal Scheduling Strategy of Microgrid Distribution Network Considering Multi-Source Agricultural Load Aggregation

Abstract

With the proposed “double carbon” target for the power system, large-scale distributed energy access poses a major challenge to the way the distribution grid operates. The rural distribution network (DN) will transform into a new local power system primarily driven by distributed renewable energy sources and energy storage, while also being interconnected with the larger power grid. The development of the rural DN will heavily rely on the construction and efficient planning of the microgrid (MG) within the agricultural park. Based on this, this paper proposes a two-stage optimal scheduling model and solution strategy for the microgrid distribution network with multi-source agricultural load aggregation. First, in the first stage, considering the flexible agricultural load and the market time-of-use electricity price, the economic optimization is realized by optimizing the operation of the schedulable resources of the park. The linear model in this stage is solved by the Lingo algorithm with fast solution speed and high accuracy. In the second stage, the power interaction between the MG and the DN in the agricultural park is considered. By optimising the output of the reactive power compensation device, the operating state of the DN is optimised. At this stage, the non-linear and convex optimization problems are solved by the particle swarm optimization algorithm. Finally, the example analysis shows that the proposed method can effectively improve the feasible region of safe operation of the distribution network in rural areas and improve the operating income of a multi-source agricultural load aggregation agricultural park.

Full text

Citation: Ma, G.; Pang, N.; Wang, Y.;
Hu, S.; Xu, X.; Zhang, Z.; Wang, C.;
Gao, L. Two-Stage Optimal
Scheduling Strategy of Microgrid
Distribution Network Considering
Multi-Source Agricultural Load
Aggregation. Energies 2024,17, 5429.
https://doi.org/10.3390/en17215429
Received: 27 September 2024
Revised: 21 October 2024
Accepted: 29 October 2024
Published: 30 October 2024
Copyright: © 2024 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
energies
Article
Two-Stage Optimal Scheduling Strategy of Microgrid
Distribution Network Considering Multi-Source Agricultural
Load Aggregation
Guozhen Ma 1, Ning Pang 1, Yunjia Wang 1, Shiyao Hu 1, Xiaobin Xu 1, Zeya Zhang 1, Changhong Wang 2
and Liai Gao 2, 3, *
1State Grid Hebei Electric Power Co., Ltd., Economic and Technological Research Institute,
Shijiazhuang 050021, China; 18003219616@189.cn (G.M.); pangning880118@hotmail.com (N.P.);
wyunjia93@126.com (Y.W.); genhsy@sina.com (S.H.); 15932135260@163.com (X.X.);
jyy_zhangzy@he.sgcc.com.cn (Z.Z.)
2College of Mechanical and Electrical Engineering, Hebei Agricultural University, Baoding 071000, China;
wangchanghong811@126.com
3
Baoding Key Laboratory of Precision Control and Clean Energy Supply for Facility Agriculture Environment,
Baoding 071000, China
*Correspondence: jdmm@hebau.edu.cn
Abstract: With the proposed “double carbon” target for the power system, large-scale distributed
energy access poses a major challenge to the way the distribution grid operates. The rural distribution
network (DN) will transform into a new local power system primarily driven by distributed renewable
energy sources and energy storage, while also being interconnected with the larger power grid. The
development of the rural DN will heavily rely on the construction and efficient planning of the
microgrid (MG) within the agricultural park. Based on this, this paper proposes a two-stage optimal
scheduling model and solution strategy for the microgrid distribution network with multi-source
agricultural load aggregation. First, in the first stage, considering the flexible agricultural load and
the market time-of-use electricity price, the economic optimization is realized by optimizing the
operation of the schedulable resources of the park. The linear model in this stage is solved by the
Lingo algorithm with fast solution speed and high accuracy. In the second stage, the power interaction
between the MG and the DN in the agricultural park is considered. By optimising the output of
the reactive power compensation device, the operating state of the DN is optimised. At this stage,
the non-linear and convex optimization problems are solved by the particle swarm optimization
algorithm. Finally, the example analysis shows that the proposed method can effectively improve the
feasible region of safe operation of the distribution network in rural areas and improve the operating
income of a multi-source agricultural load aggregation agricultural park.
Keywords: multi-source agricultural load; distribution network; optimize scheduling; reactive power
optimization; agricultural park
1. Introduction
The construction of new, rural ecological civilisations has led to a focus on rural power
system upgrading and transformation, along with the adoption of clean, economical and
efficient energy supply modes. According to statistics provided by the National Energy
Administration, by 2020, China had constructed a photovoltaic poverty-alleviation power
station with a total installed capacity of 26.36 GW. The total installed capacity of distributed
photovoltaics in China is 198 million kW, with household photovoltaics accounting for
approximately 50% of the total capacity, or 950.2 million kW. Concurrently, the National
Energy Administration proposed the construction of a number of wind power projects,
which are to be developed and utilised locally, as well as in the vicinity of villages in rural
Energies 2024,17, 5429. https://doi.org/10.3390/en17215429 https://www.mdpi.com/journal/energies

Energies 2024,17, 5429 2 of 16
areas with qualified counties. In principle, the capacity of each administrative village does
not exceed 20 MW [
1
]. This is in accordance with the national sustainable development goal
of making full use of local clean energy and of circularly transforming the existing rural
power system in accordance with local conditions [
2
]. The recently installed energy units
in rural areas are typically characterised by a relatively low capacity, a substantial quantity,
and a dispersed geographical distribution. Concurrently, the agricultural load level in
rural areas is relatively low, the load types are more diverse, and the load characteristics of
different agricultural production areas vary considerably. These factors have resulted in a
notable escalation in the financial burden associated with integrating new energy units into
the rural power grid, rendering it challenging to coordinate and oversee. The consumption
of renewable energy, such as photovoltaics, is not easily localised. The power system in
rural areas is characterised by a number of issues, including reduced voltage reliability,
power reverse transmission and the abandonment of wind and light. The promotion
of multi-source agricultural load aggregation in different agricultural production areas
and the construction of aggregated agricultural park microgrids have been identified as
effective methods for improving the flexibility of power grid dispatching management
in rural areas [
3
]. The study of the operation optimisation and energy management of
agricultural park microgrids that consider multi-source agricultural load aggregation has
significant implications for the sustainable development of new rural power systems in the
future [4].
At present, advanced energy management systems and optimisation algorithms have
improved the scheduling efficiency of microgrids [
5
] and achieved more accurate load
forecasting and resource allocation. Microgrids can adjust the power consumption mode
according to real-time power demand and price signals, as well as optimise economic
benefits [
6
]. Reference [
7
] developed an optimal microgrid operation model aimed at
minimizing operational costs to achieve efficient microgrid management. Reference [
8
] in-
troduced a two-stage interactive optimization strategy for source-grid-load-storage, which
realized a seamless integration of active and reactive power coordination control. Refer-
ence [
9
] presented a multi-objective optimal scheduling model for microgrids that considers
the uncertainties in output power from wind turbines and photovoltaic cells, facilitating
optimal microgrid scheduling. Reference [
10
] proposed a novel method for the collab-
orative optimization of rural energy systems, significantly enhancing energy efficiency.
Reference [
11
] investigated the economic dispatch problem in islanded microgrids. Refer-
ence [
12
] introduced an optimal power-scheduling framework for microgrids equipped
with integrated batteries capable of functioning in both grid-connected and islanded modes.
Although the models established in the above literature for optimal operation of microgrids
cover different optimization objectives and strategies, they fail to solve the problem of
comprehensive coordination and optimization of distributed power sources in regional
microgrids, especially in rural areas. Distributed energy sources have small capacities and
scattered locations, large variations in agricultural loads, and more flexible loads.
The integration of the microgrid in the agricultural park has added a large amount of
reactive power resources to the power system in rural areas. However, in the field of reactive
power optimisation of the distribution network, most of the reactive power compensation
devices, such as dynamic reactive power compensation devices and adjustable capacitors
with high cost, are now configured in the distribution network structure to improve the
regulation of reactive power [
13
]. Reference [
14
] optimised the reactive power output of
distributed generation to cope with the dynamic changes of the power system and improve
voltage stability. In reference [
15
], the coordinated design of shunt capacitor banks and
static var generators was carried out to solve the problem of voltage overshoot after the
integration of photovoltaics into the distribution network. Reference [
16
] addressed the
voltage issues in the distribution network by optimizing the reactive power of distributed
energy sources. Reference [
17
] focused on the economic and technical management of
distributed energy resources while optimizing power losses in the DN. Reference [
18
]
investigated the optimal reactive power compensation problem aimed at minimizing

Energies 2024,17, 5429 3 of 16
power distribution losses in smart microgrids. Reference [
19
] proposed a reactive power
optimization method based on three-stage relaxation weight and correction (TSRWC).
The above literature only focuses on the reactive power output capacity of the reactive
power compensation device, but the agricultural park microgrid with agricultural load
aggregation integrates more reactive power resources to participate in the scheduling.
Previous studies have not considered this part of the reactive power resources. It is a
problem to be solved to study how to effectively manage reactive power and further
improve the power quality of the DN in the agricultural park microgrid with new energy,
such as wind power and photovoltaic power [20].
According to the above research status and the specific needs of power system construc-
tion and development in rural areas, this paper proposes a two-stage optimal scheduling
strategy of a microgrid distribution network considering multi-source agricultural load
aggregation. In the first stage, the real-time scenarios of different production modes in
rural areas are considered, and the parks with the same production mode are aggregated
into an agricultural park microgrid that can be dispatched as a whole. According to the
day-ahead time-of-use electricity price and unit output, the ESS and demand-side resources
in the agricultural park are coordinated to optimise the operating status of the equipment
in the park and the controllable agricultural flexible load reduction, and the reactive power
capacity is reserved to respond to the reactive power demand of the system. In the second
stage, under the condition of satisfying the power flow constraints, the reactive power
resources of the microgrid in the agricultural park are fully dispatched, and the real-time
reactive power optimisation of the rural distribution network is realised by adjusting the
reactive power output of each unit and the reactive power compensation device in the
microgrid in the agricultural park. The simulation results show that the optimal scheduling
strategy can meet the requirements of the agricultural parks to improve their own income,
improve the power quality of the DN in rural areas, and reduce the network loss.
2. Agricultural Park Microgrid Structure and Optimal Scheduling Method Architecture
2.1. Microgrid Structure of Agricultural Park with Multi-Source Agricultural Load Aggregation
In the new rural power system, the agricultural park microgrid with multi-source
agricultural load aggregation contains a variety of dispatchable resources, mainly including
wind power (WT), photovoltaic (PV), biogas units (GAS), energy storage units (ESUs),
agricultural flexible load, conventional power load, and static var generators (SVGs), as well
as other components [
21
]. These components are integrated with corresponding sensors,
controllers, intelligent switches, and power converters to form controllable component
units. All the component units in the microgrid of the agricultural park are interconnected
through the communication network to facilitate the aggregation of various new energy
power generation equipment and agricultural loads in the agricultural park, enabling local
decision making and collaborative optimisation [
22
]. The MG in the agricultural park is
connected to the DN by means of a transformer situated in the station area. This enables the
distribution network to issue dispatching instructions in real time, which are then received
by the MG.
Each unit within the MG of the agricultural park is equipped with schedulable reactive
power resources. It is uncommon for the PV sets in the MG of the agricultural park to
operate at full capacity. The control of the grid-connected inverter enables the provision
of active power to the grid while simultaneously supplying the requisite reactive power.
The rotor of the WT can be controlled by the converter to provide AC excitation control,
thereby enabling the unit to emit or absorb reactive power within its capacity range in
accordance with the system scheduling. The GSA is capable of adjusting the excitation
voltage in order to provide a specified amount of reactive power to the grid. Concurrently,
SVG can be integrated with the units in the agricultural park to ensure the secure and
dependable provision of power to the DN [
23
]. Figure 1illustrates the microgrid structure
of an agricultural park with multi-source agricultural load aggregation.

Energies 2024,17, 5429 4 of 16
Energies 2024, 17, 5429 4 of 16
SVG can be integrated with the units in the agricultural park to ensure the secure and
dependable provision of power to the DN [23]. Figure 1 illustrates the microgrid structure
of an agricultural park with multi-source agricultural load aggregation.
Load
Energy Flow
Information Flow
Interactive Power
Transformer
SVG
Operating
Plan
G1
ESU
G2 G3
Load Load Load
Agricultural Park
Photovoltaic
Units Biogas Units Wind Turbines
Energy-storage
Unit
Figure 1. The microgrid structure of agricultural park with multi-source agricultural load aggregation.
2.2. Optimized Scheduling Method Architecture
This paper presents a two-stage optimal scheduling strategy of a microgrid distribu-
tion network based on the theory of multi-source agricultural load aggregation, as illus-
trated in Figure 2. In the initial phase, an MG for an agricultural park is proposed as a
means of providing energy to rural areas with diering agricultural production functions.
The objective is to establish an optimal scheduling strategy that considers the “source-
network-load-storage” interactive operation mechanism of the agricultural load aggrega-
tion agricultural park, with the goal of achieving the lowest operating cost. By optimising
the operational status of each piece of equipment in the agricultural park, the optimal
operational scheme for unit output and agricultural load adjustment in the park is ob-
tained, which promotes local energy consumption in the agricultural park and the opera-
tional income of the agricultural park. In the second stage, the DN will utilise the MG of
the agricultural park to its fullest potential in terms of reactive power resources. By ad-
justing the reactive power output of each power generation unit and reactive power com-
pensation device in the agricultural park, reactive power support for the DN in rural areas
will be provided, the voltage quality of the DN in rural areas will be enhanced, network
loss will be reduced, and the feasible region for the safe operation of the power system in
rural areas will be broadened.
Figure 1. The microgrid structure of agricultural park with multi-source agricultural load aggregation.
2.2. Optimized Scheduling Method Architecture
This paper presents a two-stage optimal scheduling strategy of a microgrid distribution
network based on the theory of multi-source agricultural load aggregation, as illustrated in
Figure 2. In the initial phase, an MG for an agricultural park is proposed as a means of pro-
viding energy to rural areas with differing agricultural production functions. The objective
is to establish an optimal scheduling strategy that considers the “source-network-load-
storage” interactive operation mechanism of the agricultural load aggregation agricultural
park, with the goal of achieving the lowest operating cost. By optimising the operational sta-
tus of each piece of equipment in the agricultural park, the optimal operational scheme for
unit output and agricultural load adjustment in the park is obtained, which promotes local
energy consumption in the agricultural park and the operational income of the agricultural
park. In the second stage, the DN will utilise the MG of the agricultural park to its fullest
potential in terms of reactive power resources. By adjusting the reactive power output of
each power generation unit and reactive power compensation device in the agricultural
park, reactive power support for the DN in rural areas will be provided, the voltage quality
of the DN in rural areas will be enhanced, network loss will be reduced, and the feasible
region for the safe operation of the power system in rural areas will be broadened.
Energies 2024, 17, 5429 5 of 16
WT
Other electricity users
PV GAS
Body: MG in agricultural park
Objective : Minimum daily operating cost
Control Variables : Generation of the unit, interruptible load, charge/
discharge of ESU, etc.
Body : Rural power DN
Objective: Minimize
network losses
Control Variables: Reactive
power output of
generators/
compensation devices, etc.
Load adjustment
Power adjustment
Interactive Power Operating Plan
ESU
Load
First : Agricultural park MG scheduling
Second: DN reactive power optimization
Pastoral Greenhouse Inhabitant
Figure 2. Two-stage optimal scheduling method architecture.
3. Mathematical Model
3.1. The First-Stage Optimal Scheduling Mathematical Model
3.1.1. First-Stage Objective Function
Considering the complementarity of generating units and the load characteristics of
dierent agricultural parks, the mathematical model is established with the goal of mini-
mizing the grid-connected operation cost of the agricultural park microgrid with multi-
source agricultural load aggregation, and with the unit output, purchase/sale of electric-
ity, baery charge and discharge, and agricultural exible load as variables [24]. The eco-
nomic operation optimization objective function of a multi-source agricultural load aggre-
gation MG in an agricultural park is as follows:

  
==
= + + + +
−
+
+ + 
 

,
cost WT WT , PV PV, GAS GAS , IL ,
11
BAT
BAT , , b b, s s ,
min [(ρ ρ ρ ρ
ρ(1 )ρ ρ ]
mt
NT IL
day t t t m t
n t M
x t x t t t
X
f C P P P P
P P P t
(1)
In the formula: Cday is the agricultural park investment costs; N represents the number
of agricultural parks; T denotes that there are T scheduling periods in a scheduling period;
∆t is the time interval; PWT,t, PPV,t, PGAS,t, and PBATx,t are the active power output of WT, PV,
GAS, and ESU in the t period, respectively;
,xt

refers to the status of the x-th distributed
energy storage device; PILm,t is the agricultural exible load;
,mt

refers to the status of the
m-th agricultural exible load equipment; ρb,t and ρs,t are the time-of-use electricity price
of purchase/sale electric energy in the t period; Pb,t and Ps,t are the electricity purchase/sale
energy in the t period; ρWT, ρPV, ρGAS, and ρBAT are the maintenance costs of WT, PV, GAS,
and ESU, respectively; ρIL is the compensation price for the interruption of the agricultural
exible load; and α is the state of purchasing/selling electricity. The value of α can only be
1 or 0, and when α is 1, which indicates that the microgrid sells electricity; when α is 0, it
means that the microgrid purchases electric energy. minfcost is the operating cost of the
agricultural park within one day.
Figure 2. Two-stage optimal scheduling method architecture.

Energies 2024,17, 5429 5 of 16
3. Mathematical Model
3.1. The First-Stage Optimal Scheduling Mathematical Model
3.1.1. First-Stage Objective Function
Considering the complementarity of generating units and the load characteristics
of different agricultural parks, the mathematical model is established with the goal of
minimizing the grid-connected operation cost of the agricultural park microgrid with
multi-source agricultural load aggregation, and with the unit output, purchase/sale of
electricity, battery charge and discharge, and agricultural flexible load as variables [
24
].
The economic operation optimization objective function of a multi-source agricultural load
aggregation MG in an agricultural park is as follows:
min fcos t =Cday +N
∑
n=1
T
∑
t=1
[(ρWT PWT,t+ρPVPPV,t+ρGASPGAS,t+ρIL∑
M
φm,tPIL
m,t+
ρBAT∑
X
ψx,tPBAT
x,t+ (1−α)ρbPb,t+αρsPs,t]∆t(1)
In the formula: C
day
is the agricultural park investment costs; Nrepresents the number
of agricultural parks; Tdenotes that there are Tscheduling periods in a scheduling period;
∆
tis the time interval; P
WT,t
,P
PV,t
,P
GAS,t
, and P
BATx,t
are the active power output of WT,
PV, GAS, and ESU in the tperiod, respectively;
ψx,t
refers to the status of the x-th distributed
energy storage device; P
ILm,t
is the agricultural flexible load;
φm,t
refers to the status of the
m-th agricultural flexible load equipment;
ρ
b,t and
ρ
s,t are the time-of-use electricity price
of purchase/sale electric energy in the tperiod; P
b,t
and P
s,t
are the electricity purchase/sale
energy in the tperiod;
ρWT
,
ρPV
,
ρGAS
, and
ρBAT
are the maintenance costs of WT, PV, GAS,
and ESU, respectively;
ρIL
is the compensation price for the interruption of the agricultural
flexible load; and
α
is the state of purchasing/selling electricity. The value of
α
can only
be 1 or 0, and when
α
is 1, which indicates that the microgrid sells electricity; when
α
is 0,
it means that the microgrid purchases electric energy. minf
cost
is the operating cost of the
agricultural park within one day.
3.1.2. Constraints
1. Active power balance constraint
PLAOD,t=PWT,t+PPV,t+PGAS,t+Pb,t+PBAT,t−Ps,t+PIL,t(2)
In the formula, PLAOD,tis the load value of the tperiod.
2. Output constraints of photovoltaic, wind, and biogas units:



PGAS,tmin ≤PGAS,t≤PGAS,tmax
PWT,tmin ≤PWT,t≤PWT,tmax
PPV,tmin ≤PPV,t≤PPV,tmax
(3)
In the formula, P
GAS,tmax
,P
GAS,tmin
,P
WT,tmin
,P
WT,tmax
,P
PV,tmin
, and P
PV,tmax
, respec-
tively, represent the maximum and minimum output levels of the GAS, WT, and PV power
generation units in the tperiod.
3. Agricultural flexible load constraints:
0≤PIL,t≤PIL,tmax (4)
In the formula, PIL,tmax is the maximum interruptible agricultural flexible load.
4. Charging and discharging constraints of ESU:

Energies 2024,17, 5429 6 of 16
A lithium battery is selected as the ESU. The charge of the lithium battery in the t
period is related to the capacity at the end of the previous period and the charge and
discharge power in this period; that is:
EBAT,t=EBAT,t−1(1−σ)−∆T×θdown,t ×Pdown
βdown
+∆T×θup,tPup ×βup (5)
In the formula, E
BAT,t
is the charge of the lithium battery in the tperiod;
σ
is the
self-discharge rate of the lithium battery, where this paper takes 0; and P
up,t
,P
down,t
,
βup
,
and
βdown
are the charge/discharge power and charge/discharge efficiency of the lithium
battery in the t period.
θup,t
,
θdown,t
are the charging/discharging state marks of the lithium
battery in a scheduling cycle, and the values of the two are only 1 and 0. Taking 1 indicates
that the lithium battery is in the charging and discharging state, and taking 0 indicates that
the lithium battery is not in the charging and discharging state.
5. State of charge constraints of ESU:
In a scheduling cycle, the lithium battery should ensure that the state of charge is
equal at the beginning and end of a scheduling cycle. That is:
EBAT,min ≤EBAT,t≤EBAT,max (6)
EBAT,t=EBAT,INIT (7)
In the formula: E
BAT,min
and E
BAT,max
represent the minimum/maximum charge of
the lithium battery, respectively. E
BAT,t
represents the charge of the energy storage device in
the tperiod; E
BAT,INIT
represents the charge of the energy storage device in the initial state.
6. Charging and discharging power constraint:
Pdown,tmin ≤Pdown,t≤Pdown,tmax
Pup,tmin ≤Pup,t≤Pup,tmax (8)
In the formula, P
up,tmax
,P
down,tmax
,P
up,tmin
, and P
down,tmin
are the upper and lower
limits of the charging and discharging power of the battery in the t period, respectively.
7. Uniqueness constraint of battery charge and discharge state:
θup,t+θdown,t≤1 (9)
In the formula, the battery can only be charged and discharged simultaneously.
3.2. The Second-Stage Optimal Scheduling Mathematical Model
In the second stage of reactive power optimisation of DN, it is essential to consider
the specific structure of rural DN and the grid-connected position of each MG and reactive
power compensation device, as well as to optimise the reactive power output of each
microgrid and reactive power compensation device [
25
]. The objective of the second stage
of optimisation is to minimise the active loss of the rural DN while introducing a penalty
function, for instance, where the node voltage exceeds the specified limit [
26
]. The decision
variables include the reactive power output of the MG in each agricultural park and the
output from the reactive power compensation equipment
3.2.1. Second-Stage Objective Function
The second-stage optimization objective function:
Closse =min[Plosse +λ
n
∑
i=1
(∆Ui
Ui,max −Ui,min
)
2
](10)
∆Ui=


Ui,min −Ui,Ui<Ui,min
0, Ui,min ≤Ui≤Ui,max
Ui−Ui,max,Ui>Ui,max
(11)

Energies 2024,17, 5429 7 of 16
Plosse =
N
∑
n=1
T
∑
t=1
I2
nRn(12)
In the formula, P
losse
is the network loss power of the DN in a scheduling period of
the rural DN; ndenotes the number of nodes in rural areas; and U
i
,U
i,min
, and U
i,max
are
the voltage of the rural DN at node iand its allowable upper and lower limits of voltage,
respectively. N is the number of network branches; I
n
in is the branch current value of the
nth branch; R
n
is the resistance value of the nth branch; and
λ
is the penalty coefficient of
the voltage over-limit.
3.2.2. Constraint Conditions
1. Power flow constraints.







PGi +PDi +PMi −PLi =Ui
n
∑
j=1
Uj(Gij cos θij +Bij sin θij)
QGi +QDi +QMi −QLi =Ui
n
∑
j=1
Uj(Gij sin θij −Bij cos θij)
(13)
In the formula, the variables P
Gi
,P
Di
, and P
Mi
represent the active power injected into
node (i) of the main network, distributed generation, and MG, respectively. Meanwhile,
Q
Gi
,Q
Di
, and Q
Mi
denote the reactive power injected into node (i). P
Li
and Q
Li
refer to the
active and reactive loads at node (i), respectively. U
i
and U
j
are the voltage amplitudes
of nodes (i) and (j), respectively. Additionally, G
ij
,B
ij
and
θij
represent the conductance,
susceptance, and phase difference between nodes (i) and (j) in the DN, respectively.
2. Node voltage constraint:
Ui,min ≤Ui≤Ui,max (14)
In the formula, U
i,min
,U
i,max
are the minimum and maximum values of the allowable
voltage of the distribution network node i.
3. Reactive power compensation constraints of a microgrid in an agricultural park:
QMG
i,tmin ≤QMG
i,t≤QMG
i,tmax (15)
In the formula, Q
MGi,tmin
and Q
MGi,tmax
are the upper and lower limits of the reactive
power output of the microgrid with the grid-connected node at iat time t, respectively.
4. Capacity constraint of reactive power compensator:
QC
i,min ≤QC
i≤QC
i,max (16)
In the formula, Q
Ci,min
,Q
Ci,max
are the upper and lower limits of the capacity of the
reactive power compensation device installed at node i, respectively.
5. Tie-line power constraint:
Ppcc_min ≤Ppcc,t≤Ppcc_max
Qpcc_min ≤Qpcc,t≤Qpcc_max (17)
In the formula, P
pcc,t
is the interactive active power between the microgrid and the
distribution network; Q
pcc,t
is the interactive active power and reactive power of the
microgrid and the distribution network; P
pcc_min
and P
pcc_max
are the lower and upper
limits of the interactive active power, respectively; and Q
pcc_min
and Q
pcc_max
are the lower
and upper limits of the interactive reactive power.
4. Mathematical Model Solving Method
The Lingo 18.0 optimisation software is designed to address a range of optimisation
problems. The software is capable of rapidly and accurately identifying the optimal solution

Energies 2024,17, 5429 8 of 16
or an approximate optimal solution to the problem. In the mathematical model of MG grid-
connected optimal scheduling in agricultural parks, the Lingo 18.0 software is employed for
the initial data programming and model solution, thereby obtaining the optimal solution.
The reactive power optimization problem in a DN with a microgrid can be framed as
a multi-dimensional nonlinear optimization challenge. This mathematical model involves
two key components: calculating power flow and optimizing reactive power output within
the DN [
27
]. To solve the power flow model, the MATLAB-based simulation toolkit
MATPOWER 7.1 is utilized, allowing for the determination of node voltages and network
loss values through the power flow calculations. The particle swarm optimization algorithm
is then used to optimize the reactive power output of MG and compensation device over
time by iteratively updating each generation of particles. The process for solving the
mathematical model is illustrated in Figure 3. Each rectangular box in the flowchart
represents the solution step, the arrow indicates the direction of the solution, and the
diamond box represents the condition determination.
Energies 2024, 17, 5429 9 of 16
Start
Develop a cost-optimal Stage 1 economic dispatch model
for multiple agricultural park systems.
Determine the decision variables
Output the Stage 1 optimization results
Determine the interaction power between the MG and
the DN based on the optimization results
Solving mathematical models using LINGO software
Formulate the Stage 2 optimization scheduling model
for minimizing network losses in the DN
Determine the maximum adjustable range of
reactive power output for the MG.
Output of MG and compensation devices for
reactive power on an hourly basis
less than the maximum allowable power
flow deviation
power flow calculation
distribution network loss optimization
END
Input the parameters for the DN and set initial values
Calculate the fitness value of the particles
Calculate the historical best positions and
optimal fitness values of the population
Use PSO to solve the Stage 2 model
Update the particles' velocities and positions.
Determine whether the termination
criteria are met
NO
NO
YES
YES
Figure 3. Mathematical model solving process.
5. Example Analysis
5.1. Example Description
In rural areas, three typical gathering areas can be identied: those associated with
villages, animal husbandry, and greenhouses. This example presents a microgrid model
constructed based on the aforementioned three types of agricultural parks. Once the agri-
cultural park’s microgrid is connected to the main grid, it will be able to buy or sell elec-
tricity within the DN. The PV output data are derived from the PV output of a greenhouse
on a day with optimal solar irradiance. The WT power output is based on the output of a
WT on a specic day. The GAS output data are based on the power generation output of
a cogeneration unit on a specic day. Four LFP 5000/HV LiFePO4 baeries were selected
to form a set of ESU, and the parameters of a single baery are presented in Table 1. The
parameters of the microgrid system and the maintenance costs of each unit in the agricul-
tural park [28,29] are shown in Table 2.
Table 1. Single energy storage baery parameters.
Charge/Discharge
Eciency (kw/h)
Upper Limit of
Charging/Dis-
charging Power
(kw/h)
Lower Limit of
Charge/Discharge
Power (kw/h)
Lower Limit of
Energy Storage
(kwh)
Initial Energy
Storage (kwh)
Total Energy
(kwh)
0.9
2.5
0
0.51
1.5
5.12
Table 2. Three types of agricultural park microgrid system parameters.
Microgrid Type
Unit Type
Installed Capacity
Installation Node
SVG Capacity
Maintenance Cost
Number of Clas-
ses
Greenhouse
MG1
WT
1.5 MW
7
0.2 Mvar
0.11/¥
1
PV
12 kW
0.08/¥
100
ESU
20 kwh
0.3/¥
10
Pastoral
MG2
PV
1 kW
24
0.2 Mvar
0.08/¥
100
GAS
1.5 MW
0.34/¥
1
Inhabitant
MG3
PV
12 kW
25
0.2 Mvar
0.08/¥
100
GAS
1.2 MW
0.34/¥
1
Figure 3. Mathematical model solving process.
5. Example Analysis
5.1. Example Description
In rural areas, three typical gathering areas can be identified: those associated with
villages, animal husbandry, and greenhouses. This example presents a microgrid model
constructed based on the aforementioned three types of agricultural parks. Once the
agricultural park’s microgrid is connected to the main grid, it will be able to buy or
sell electricity within the DN. The PV output data are derived from the PV output of a
greenhouse on a day with optimal solar irradiance. The WT power output is based on the
output of a WT on a specific day. The GAS output data are based on the power generation
output of a cogeneration unit on a specific day. Four LFP 5000/HV LiFePO4 batteries
were selected to form a set of ESU, and the parameters of a single battery are presented in
Table 1. The parameters of the microgrid system and the maintenance costs of each unit in
the agricultural park [28,29] are shown in Table 2.

Energies 2024,17, 5429 9 of 16
Table 1. Single energy storage battery parameters.
Charge/Discharge
Efficiency (kw/h)
Upper Limit of
Charging/Discharging
Power (kw/h)
Lower Limit of
Charge/Discharge
Power (kw/h)
Lower Limit of
Energy Storage
(kwh)
Initial Energy
Storage (kwh)
Total Energy
(kwh)
0.9 2.5 0 0.51 1.5 5.12
Table 2. Three types of agricultural park microgrid system parameters.
Microgrid
Type Unit Type Installed
Capacity
Installation
Node
SVG
Capacity
Maintenance
Cost
Number of
Classes
Greenhouse
MG1
WT 1.5 MW
70.2 Mvar
0.11/¥ 1
PV 12 kW 0.08/¥ 100
ESU 20 kwh 0.3/¥ 10
Pastoral
MG2
PV 1 kW 24 0.2 Mvar 0.08/¥ 100
GAS 1.5 MW 0.34/¥ 1
Inhabitant
MG3
PV 12 kW 25 0.2 Mvar 0.08/¥ 100
GAS 1.2 MW 0.34/¥ 1
The reference voltage of the DN is 10 kilovolts (kV). The total network load of the
standard IEEE 33-bus distribution system is 5084.26 + j2547.32 kVA. The IEEE33 node DN
example has been enhanced, with node 1 designated as the transmission and distribution
contact node. It is further assumed that three groups of switchable parallel capacitors (QC1,
QC2, QC3) are connected in parallel at distribution network nodes 8, 30, and 32, respectively,
with each group having a capacity of 800 kvar [
30
]. The remaining nodes represent ordinary
user loads. In light of the aforementioned details, the distribution network structure has
been enhanced on the basis of the IEEE 33 example network structure, forming the rural
distribution network system structure with the agricultural park micro-grid, as illustrated
in Figure 4.
Energies 2024, 17, 5429 10 of 16
The reference voltage of the DN is 10 kilovolts (kV). The total network load of the
standard IEEE 33-bus distribution system is 5084.26 + j2547.32 kVA. The IEEE33 node DN
example has been enhanced, with node 1 designated as the transmission and distribution
contact node. It is further assumed that three groups of switchable parallel capacitors
(QC1, QC2, QC3) are connected in parallel at distribution network nodes 8, 30, and 32,
respectively, with each group having a capacity of 800 kvar [30]. The remaining nodes
represent ordinary user loads. In light of the aforementioned details, the distribution net-
work structure has been enhanced on the basis of the IEEE 33 example network structure,
forming the rural distribution network system structure with the agricultural park micro-
grid, as illustrated in Figure 4.
1
2
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
19 20 21 22
23 24 25 26 27 28 29 30 31 32 33
MG1
MG3
MG2
QC1
QC2QC3
Figure 4. Improved IEEE33 node example diagram.
To account for the varying loads during dierent periods within a scheduling cycle,
this paper uses the total load from the IEEE33 standard example as a benchmark. It is
assumed that the ratio of each node’s active load in the DN to the system’s total active
load remains constant. The active and reactive loads at other times are then calculated
using the following formula:
N
1
'j,t t j j
j
P P P P
=
=

(18)
tan(arccos( ))
j,t j j,t
Q' P'

=
(19)
In the formula, N is the total number of nodes in the distribution network;
'j,t
P
is
the actual active load incorporated into the DN j node at time t; Pj is the load of node j in
the standard example; Pt represents the actual total active load of the DN at time t;
j,t
Q'
is the reactive load incorporated into the j node of the DN at time t; and φj is the load
power factor at node j.
5.2. Analysis of Optimal Scheduling Results
After the optimized scheduling, the output of each unit in the microgrids of the var-
ious parks, along with the ESU charge level and charging/discharging power, the magni-
tude of interruptible agricultural exible loads, and the electricity purchased from the
grid, as well as the net revenue and generation costs for each park over a day, are illus-
trated in Figures 5, Figure 6, and Figure 7, respectively.
The greenhouse park requires the rolling machine to be operational at 07:00 and
16:00. The time is 00:00. In the greenhouse where strawberries are cultivated, in addition
to the electrical equipment that requires continuous operation, the ventilator is activated
when the relative humidity exceeds 81% RH. During the day, the temperature is main-
tained below 16 °C to enable the opening of the heater. At night, the heater is activated
when the temperature drops below 4 °C. The daytime photosynthetic active radiation
value is below 340 μmol/m²/s to trigger the illumination of the greenhouse [31,32]. In the
greenhouse park, the highest demand for electricity is observed between 7:00 and 9:00, as
Figure 4. Improved IEEE33 node example diagram.
To account for the varying loads during different periods within a scheduling cycle,
this paper uses the total load from the IEEE33 standard example as a benchmark. It is
assumed that the ratio of each node’s active load in the DN to the system’s total active load
remains constant. The active and reactive loads at other times are then calculated using the
following formula:
P′j,t=Pt,N
∑
j=1
Pj×Pj(18)
Q′j,t=tan(arccos(ϕj)) ×P′j,t(19)
In the formula, N is the total number of nodes in the distribution network;
P′j,t
is the
actual active load incorporated into the DN jnode at time t;P
j
is the load of node jin the
standard example; P
t
represents the actual total active load of the DN at time t;
Q′j,t
is the
reactive load incorporated into the jnode of the DN at time t; and
φj
is the load power
factor at node j.

Energies 2024,17, 5429 10 of 16
5.2. Analysis of Optimal Scheduling Results
After the optimized scheduling, the output of each unit in the microgrids of the various
parks, along with the ESU charge level and charging/discharging power, the magnitude of
interruptible agricultural flexible loads, and the electricity purchased from the grid, as well
as the net revenue and generation costs for each park over a day, are illustrated in Figure 5,
Figure 6, and Figure 7, respectively.
Energies 2024, 17, 5429 11 of 16
well as between 16:00 and 17:00. The period between 07:00 and 17:00 is characterised by a
peak in electricity consumption, while the remainder of the day is associated with a trough
in consumption. Following the scheduling of the aforementioned processes, at 08:00 the
following morning, the greenhouse park is no longer able to purchase electricity from the
grid. This is due to the utilisation of energy storage devices, which facilitate the provision
of power and the interruption of agricultural loads. Conversely, at 23:00–24:00, the valley
period of electricity consumption occurs. This coincides with a reduction in the cost of
purchasing electricity. The greenhouse park, therefore, purchases electricity from the grid.
Figure 5. The results of optimal scheduling of microgrid in greenhouse park.
Figure 6. Results of optimal dispatching of microgrid in residential park.
The electricity load of the residential park is relatively stable, with only a slight in-
crease at 08:00 and 19:00. The output of the unit in the residential park is sucient to meet
the park’s load without the need to purchase electricity from the distribution network.
Furthermore, any surplus electricity can be sold to the distribution network. The residen-
tial park microgrid is integrated into the distribution network as a power source within a
dened scheduling cycle. The energy storage device is not in operation, and its charge
remains unaltered.
It is imperative that the animal husbandry park activates the fan and wet curtain
cooling ventilation for an extended period between the hours of 6:00 and 19:00. This will
result in a signicant surge in electricity consumption during this designated time frame.
The animal husbandry park procures electricity from the power grid for the purpose of
0 2 4 6 8 10 12 14 16 18 20 22 24
0
50
100
150
200
250
300 Cost
Net Yield
ESU Charge
ESU Charging
Load
ESU Discharge
Purchase
IL
WT
PV
Time/h
Active Power/kw
-100
0
100
200
300
400
500
Yield/¥
0 2 4 6 8 10 12 14 16 18 20 22 24
0
50
100
150
200
250
300
350
400 Cost
Net Yield
ESU Charge
ESU Charging
Load
GAS
PV
Time/h
Active Power/kw
-150
-100
-50
0
50
100
150
200
250
300
350
Yield/¥
Figure 5. The results of optimal scheduling of microgrid in greenhouse park.
Energies 2024, 17, 5429 11 of 16
well as between 16:00 and 17:00. The period between 07:00 and 17:00 is characterised by a
peak in electricity consumption, while the remainder of the day is associated with a trough
in consumption. Following the scheduling of the aforementioned processes, at 08:00 the
following morning, the greenhouse park is no longer able to purchase electricity from the
grid. This is due to the utilisation of energy storage devices, which facilitate the provision
of power and the interruption of agricultural loads. Conversely, at 23:00–24:00, the valley
period of electricity consumption occurs. This coincides with a reduction in the cost of
purchasing electricity. The greenhouse park, therefore, purchases electricity from the grid.
Figure 5. The results of optimal scheduling of microgrid in greenhouse park.
Figure 6. Results of optimal dispatching of microgrid in residential park.
The electricity load of the residential park is relatively stable, with only a slight in-
crease at 08:00 and 19:00. The output of the unit in the residential park is sucient to meet
the park’s load without the need to purchase electricity from the distribution network.
Furthermore, any surplus electricity can be sold to the distribution network. The residen-
tial park microgrid is integrated into the distribution network as a power source within a
dened scheduling cycle. The energy storage device is not in operation, and its charge
remains unaltered.
It is imperative that the animal husbandry park activates the fan and wet curtain
cooling ventilation for an extended period between the hours of 6:00 and 19:00. This will
result in a signicant surge in electricity consumption during this designated time frame.
The animal husbandry park procures electricity from the power grid for the purpose of
0 2 4 6 8 10 12 14 16 18 20 22 24
0
50
100
150
200
250
300 Cost
Net Yield
ESU Charge
ESU Charging
Load
ESU Discharge
Purchase
IL
WT
PV
Time/h
Active Power/kw
-100
0
100
200
300
400
500
Yield/¥
0 2 4 6 8 10 12 14 16 18 20 22 24
0
50
100
150
200
250
300
350
400 Cost
Net Yield
ESU Charge
ESU Charging
Load
GAS
PV
Time/h
Active Power/kw
-150
-100
-50
0
50
100
150
200
250
300
350
Yield/¥
Figure 6. Results of optimal dispatching of microgrid in residential park.
Energies 2024, 17, 5429 12 of 16
charging the ESU between the hours of 00:00 and 08:00, as well as between 18:00 and 18:00.
From 00:00 to 08:00, the park experiences a decit in electricity, necessitating the utilisa-
tion of the ESU to meet the park’s energy demands. At 11:00, 13:00, and 17:00, the unit output
is sufficient to charge the energy storage system, which is then powered when the animal hus-
bandry park is short of electricity at 18:00. The period is between 00:00 and 19:00.
Figure 7. The results of optimal scheduling of microgrid in animal husbandry park.
Following the initial phase of collaborative optimisation scheduling of the aforemen-
tioned three types of agricultural park microgrids, the agricultural park is in a position to
sell the surplus electricity to the power grid, thereby obtaining benets [33]. This signi-
cantly enhances the operational eciency of the agricultural park microgrid. Table 3 illus-
trates the comparative net income of each microgrid following the optimal scheduling of
grid-connected operation.
Table 3. Comparison of net income of microgrid in dierent agricultural park scenarios.
Microgrid Type
Do Not Install
Generator
Sets/CNY
Installation Unit is
Not Connected to
the Grid/CNY
Unoptimized Sched-
uling after Grid
Connection/CNY
Optimized Schedul-
ing after Grid Con-
nection/CNY
Greenhouse MG1
−1454
−1628.4
2647.8
2666.6
Pastoral MG2
−4797.9
−1920.4
−823.7
−551.6
Inhabitant MG3
−2624.6
−710.8
1123.4
1123.4
The interactive power between the microgrid of the agricultural park composed of
each unit, energy storage device, and agricultural load and the distribution network after
the rst stage of optimal scheduling is shown in Figure 8.
Taking the microgrid of the animal husbandry park as an example, when Ppcc,t ≤ 0, the
greenhouse as a load does not have a voltage regulation eect; when Ppcc,t ≥ 0, it means
that the microgrid in the animal husbandry park is used as a power source, and the mi-
crogrid has a voltage regulation eect. Similarly, when Qpcc,t ≤ 0, the livestock park is in-
corporated into the distribution network as a reactive load. When the interactive reactive
power is greater than Qpcc,t ≥ 0, the livestock park can be regarded as a reactive power
compensation device, which can output reactive power.
Based on the Intel (R) Core (TM) i5-8300H processor 8G RAM computer, the simu-
lation programming is carried out in the MATLAB R2022B software environment, and the
reactive power output of each microgrid and automatic capacitor in each period of a
scheduling cycle is solved as shown in Figure 9. The units in the microgrid exhibit a higher
output at the daytime hours of 9:00–17:00, which can provide a greater reactive power
0 2 4 6 8 10 12 14 16 18 20 22 24
0
50
100
150
200
250
300
350
400 Cost
Net Yield
ESU Charge
ESU Charging
Load
ESU Discharge
Purchase
IL
GAS
PV
Time/h
Active Power/kw
-150
-100
-50
0
50
100
150
200
250
300
Yield/¥
Figure 7. The results of optimal scheduling of microgrid in animal husbandry park.

Energies 2024,17, 5429 11 of 16
The greenhouse park requires the rolling machine to be operational at 07:00 and 16:00.
The time is 00:00. In the greenhouse where strawberries are cultivated, in addition to the
electrical equipment that requires continuous operation, the ventilator is activated when
the relative humidity exceeds 81% RH. During the day, the temperature is maintained
below 16
◦
C to enable the opening of the heater. At night, the heater is activated when the
temperature drops below 4
◦
C. The daytime photosynthetic active radiation value is below
340
µ
mol/m²/s to trigger the illumination of the greenhouse [
31
,
32
]. In the greenhouse
park, the highest demand for electricity is observed between 7:00 and 9:00, as well as
between 16:00 and 17:00. The period between 07:00 and 17:00 is characterised by a peak
in electricity consumption, while the remainder of the day is associated with a trough
in consumption. Following the scheduling of the aforementioned processes, at 08:00 the
following morning, the greenhouse park is no longer able to purchase electricity from the
grid. This is due to the utilisation of energy storage devices, which facilitate the provision
of power and the interruption of agricultural loads. Conversely, at 23:00–24:00, the valley
period of electricity consumption occurs. This coincides with a reduction in the cost of
purchasing electricity. The greenhouse park, therefore, purchases electricity from the grid.
The electricity load of the residential park is relatively stable, with only a slight
increase at 08:00 and 19:00. The output of the unit in the residential park is sufficient to
meet the park’s load without the need to purchase electricity from the distribution network.
Furthermore, any surplus electricity can be sold to the distribution network. The residential
park microgrid is integrated into the distribution network as a power source within a
defined scheduling cycle. The energy storage device is not in operation, and its charge
remains unaltered.
It is imperative that the animal husbandry park activates the fan and wet curtain
cooling ventilation for an extended period between the hours of 6:00 and 19:00. This will
result in a significant surge in electricity consumption during this designated time frame.
The animal husbandry park procures electricity from the power grid for the purpose of
charging the ESU between the hours of 00:00 and 08:00, as well as between 18:00 and 18:00.
From 00:00 to 08:00, the park experiences a deficit in electricity, necessitating the utilisation
of the ESU to meet the park’s energy demands. At 11:00, 13:00, and 17:00, the unit output
is sufficient to charge the energy storage system, which is then powered when the animal
husbandry park is short of electricity at 18:00. The period is between 00:00 and 19:00.
Following the initial phase of collaborative optimisation scheduling of the aforemen-
tioned three types of agricultural park microgrids, the agricultural park is in a position
to sell the surplus electricity to the power grid, thereby obtaining benefits [
33
]. This sig-
nificantly enhances the operational efficiency of the agricultural park microgrid. Table 3
illustrates the comparative net income of each microgrid following the optimal scheduling
of grid-connected operation.
Table 3. Comparison of net income of microgrid in different agricultural park scenarios.
Microgrid Type Do Not Install
Generator Sets/CNY
Installation Unit is Not
Connected to the
Grid/CNY
Unoptimized
Scheduling after Grid
Connection/CNY
Optimized Scheduling
after Grid
Connection/CNY
Greenhouse MG1 −1454 −1628.4 2647.8 2666.6
Pastoral MG2 −4797.9 −1920.4 −823.7 −551.6
Inhabitant MG3 −2624.6 −710.8 1123.4 1123.4
The interactive power between the microgrid of the agricultural park composed of
each unit, energy storage device, and agricultural load and the distribution network after
the first stage of optimal scheduling is shown in Figure 8.

Energies 2024,17, 5429 12 of 16
Energies 2024, 17, 5429 13 of 16
output for the DN, thereby optimising the distribution network loss. In the event that the
unit in the MG is unable to provide an adequate reactive power output at 18, from 8:00
p.m. to 8:00 a.m., the MG is reliant on SVG and capacitor banks in other DN stations for
reactive power output and distribution network loss optimisation.
Figure 8. The interactive active power of microgrid and distribution network in each agricultural park.
Figure 9. The reactive power output of microgrid and reactive power compensation device in each
agricultural park.
After the microgrid in the agricultural park is connected to the grid, the reactive
power output of each microgrid and compensation device is optimized within their re-
spective reactive power output ranges. The optimized reactive power output is then ap-
plied to an improved IEEE33 distribution network case study using the MATPOWER
power system simulation optimization toolbox to solve the distribution network’s power
ow. The comparison of node voltage magnitudes before and after optimization during
various time periods is illustrated in Figure 10. Throughout the entire scheduling period,
from 10:00 to 14:00, the MG delivers a substantial amount of active power to the DN, re-
sulting in a signicant voltage increase during this time frame. Furthermore, after the two-
stage optimization scheduling, the voltage magnitudes at all nodes in the DN remain
above 0.9, thereby staying within the voltage constraint range.
0 2 4 6 8 10 12 14 16 18 20 22 24
-100
0
100
200
300
400
500
600
700
Interactive Power/kw
Time/h
MG1
MG2
MG3
0 2 4 6 8 10 12 14 16 18 20 22 24
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5 QC3
MG1 MG2
MG3
Reactive Power/Mvar
Time/h
QC1
QC2
Figure 8. The interactive active power of microgrid and distribution network in each agricultural park.
Taking the microgrid of the animal husbandry park as an example, when
Ppcc,t≤0,
the greenhouse as a load does not have a voltage regulation effect; when P
pcc,t≥
0, it
means that the microgrid in the animal husbandry park is used as a power source, and the
microgrid has a voltage regulation effect. Similarly, when Q
pcc,t≤
0, the livestock park is
incorporated into the distribution network as a reactive load. When the interactive reactive
power is greater than Q
pcc,t≥
0, the livestock park can be regarded as a reactive power
compensation device, which can output reactive power.
Based on the Intel (R) Core (TM) i5-8300H processor 8G RAM computer, the simulation
programming is carried out in the MATLAB R2022B software environment, and the reactive
power output of each microgrid and automatic capacitor in each period of a scheduling
cycle is solved as shown in Figure 9. The units in the microgrid exhibit a higher output
at the daytime hours of 9:00–17:00, which can provide a greater reactive power output
for the DN, thereby optimising the distribution network loss. In the event that the unit in
the MG is unable to provide an adequate reactive power output at 18, from 8:00 p.m. to
8:00 a.m., the MG is reliant on SVG and capacitor banks in other DN stations for reactive
power output and distribution network loss optimisation.
Energies 2024, 17, 5429 13 of 16
output for the DN, thereby optimising the distribution network loss. In the event that the
unit in the MG is unable to provide an adequate reactive power output at 18, from 8:00
p.m. to 8:00 a.m., the MG is reliant on SVG and capacitor banks in other DN stations for
reactive power output and distribution network loss optimisation.
Figure 8. The interactive active power of microgrid and distribution network in each agricultural park.
Figure 9. The reactive power output of microgrid and reactive power compensation device in each
agricultural park.
After the microgrid in the agricultural park is connected to the grid, the reactive
power output of each microgrid and compensation device is optimized within their re-
spective reactive power output ranges. The optimized reactive power output is then ap-
plied to an improved IEEE33 distribution network case study using the MATPOWER
power system simulation optimization toolbox to solve the distribution network’s power
ow. The comparison of node voltage magnitudes before and after optimization during
various time periods is illustrated in Figure 10. Throughout the entire scheduling period,
from 10:00 to 14:00, the MG delivers a substantial amount of active power to the DN, re-
sulting in a signicant voltage increase during this time frame. Furthermore, after the two-
stage optimization scheduling, the voltage magnitudes at all nodes in the DN remain
above 0.9, thereby staying within the voltage constraint range.
0 2 4 6 8 10 12 14 16 18 20 22 24
-100
0
100
200
300
400
500
600
700
Interactive Power/kw
Time/h
MG1
MG2
MG3
0 2 4 6 8 10 12 14 16 18 20 22 24
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5 QC3
MG1 MG2
MG3
Reactive Power/Mvar
Time/h
QC1
QC2
Figure 9. The reactive power output of microgrid and reactive power compensation device in each
agricultural park.

Energies 2024,17, 5429 13 of 16
After the microgrid in the agricultural park is connected to the grid, the reactive power
output of each microgrid and compensation device is optimized within their respective
reactive power output ranges. The optimized reactive power output is then applied to an
improved IEEE33 distribution network case study using the MATPOWER power system
simulation optimization toolbox to solve the distribution network’s power flow. The
comparison of node voltage magnitudes before and after optimization during various time
periods is illustrated in Figure 10. Throughout the entire scheduling period, from 10:00
to 14:00, the MG delivers a substantial amount of active power to the DN, resulting in
a significant voltage increase during this time frame. Furthermore, after the two-stage
optimization scheduling, the voltage magnitudes at all nodes in the DN remain above 0.9,
thereby staying within the voltage constraint range.
Energies 2024, 17, 5429 14 of 16
A comparison of the network loss at each moment before and after reactive power
compensation and the optimisation of the reactive power output of the microgrid is pre-
sented in Figure 11. Following the integration of the microgrid into the DN, a notable
reduction in active power loss within the distribution system is observed over the course
of each scheduling cycle. This is accomplished by employing a reactive power compensa-
tion device and optimizing the reactive power within the MG. Prior to reactive power
compensation, the network loss was 8.997 MW. Following the implementation of reactive
power optimization, which included the use of a reactive power compensation device and
microgrid, the network loss was reduced to 5.541 MW.
Figure 10. Comparison diagram of voltage amplitude of each node in distribution network before
and after reactive power optimization.
Figure 11. Comparison diagram of distribution network loss before and after reactive power opti-
mization.
6. Conclusions
This paper proposes a two-stage interactive optimal scheduling strategy for a mi-
crogrid distribution network that considers multi-source agricultural load aggregation.
The objective is to address the issues of power ow return and voltage over-limit that arise
when a microgrid of an agricultural park with agricultural load aggregation is integrated
into the distribution network in rural areas. The eectiveness of the proposed scheduling
strategy is demonstrated through a thorough analysis of the enhanced IEEE 33-node dis-
tribution network system. The main conclusions are as follows:
10
20
10
20
30
0.86
0.88
0.90
0.92
0.94
0.96
0.98
1.00
Hop Count
Voltage Amplitude/pu
Time/h
Before optimization
After optimization
0 2 4 6 8 10 12 14 16 18 20 22 24
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
0.50
Network Loss/MW
Time/h
Before Optimization
After Optimization
Figure 10. Comparison diagram of voltage amplitude of each node in distribution network before
and after reactive power optimization.
A comparison of the network loss at each moment before and after reactive power
compensation and the optimisation of the reactive power output of the microgrid is pre-
sented in Figure 11. Following the integration of the microgrid into the DN, a notable
reduction in active power loss within the distribution system is observed over the course
of each scheduling cycle. This is accomplished by employing a reactive power compen-
sation device and optimizing the reactive power within the MG. Prior to reactive power
compensation, the network loss was 8.997 MW. Following the implementation of reactive
power optimization, which included the use of a reactive power compensation device and
microgrid, the network loss was reduced to 5.541 MW.
Energies 2024, 17, 5429 14 of 16
A comparison of the network loss at each moment before and after reactive power
compensation and the optimisation of the reactive power output of the microgrid is pre-
sented in Figure 11. Following the integration of the microgrid into the DN, a notable
reduction in active power loss within the distribution system is observed over the course
of each scheduling cycle. This is accomplished by employing a reactive power compensa-
tion device and optimizing the reactive power within the MG. Prior to reactive power
compensation, the network loss was 8.997 MW. Following the implementation of reactive
power optimization, which included the use of a reactive power compensation device and
microgrid, the network loss was reduced to 5.541 MW.
Figure 10. Comparison diagram of voltage amplitude of each node in distribution network before
and after reactive power optimization.
Figure 11. Comparison diagram of distribution network loss before and after reactive power opti-
mization.
6. Conclusions
This paper proposes a two-stage interactive optimal scheduling strategy for a mi-
crogrid distribution network that considers multi-source agricultural load aggregation.
The objective is to address the issues of power ow return and voltage over-limit that arise
when a microgrid of an agricultural park with agricultural load aggregation is integrated
into the distribution network in rural areas. The eectiveness of the proposed scheduling
strategy is demonstrated through a thorough analysis of the enhanced IEEE 33-node dis-
tribution network system. The main conclusions are as follows:
10
20
10
20
30
0.86
0.88
0.90
0.92
0.94
0.96
0.98
1.00
Hop Count
Voltage Amplitude/pu
Time/h
Before optimization
After optimization
0 2 4 6 8 10 12 14 16 18 20 22 24
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
0.50
Network Loss/MW
Time/h
Before Optimization
After Optimization
Figure 11. Comparison diagram of distribution network loss before and after reactive power optimization.

Energies 2024,17, 5429 14 of 16
6. Conclusions
This paper proposes a two-stage interactive optimal scheduling strategy for a micro-
grid distribution network that considers multi-source agricultural load aggregation. The
objective is to address the issues of power flow return and voltage over-limit that arise when
a microgrid of an agricultural park with agricultural load aggregation is integrated into the
distribution network in rural areas. The effectiveness of the proposed scheduling strategy
is demonstrated through a thorough analysis of the enhanced IEEE 33-node distribution
network system. The main conclusions are as follows:
1.
The typical curves and time-varying characteristics of each energy consumption load
in the agricultural park are subjected to analysis. The matching between supply and
demand is enhanced through the interactive operation of the internal network load
storage in the park. The local consumption rate of new energy in the park is increased
from 47% to 72.6%, resulting in a net income increase of CNY 272.2 per day.
2.
In an electricity market environment, the proposed two-stage optimal scheduling
strategy can stabilise the voltage of the distribution network in rural areas above
0.9 pu through the optimal scheduling of the reactive power output of the microgrid
and the reactive power compensation device. It was found that, within a reasonable
range of voltage, the network loss of the DN could be reduced from 8.997 MW to 5.541
MW, representing a reduction in the network loss rate of 38.4%.
3.
The two-stage optimized scheduling method for microgrid distribution networks,
which incorporates the aggregation of multi-source agricultural loads proposed in
this paper, can serve as an effective reference for practical power networks in rural
areas. Furthermore, collaboration and complementarity among different distribution
areas can be considered to promote internal electricity consumption within agricul-
tural parks.
Author Contributions: G.M. and N.P.; writing—original draft, G.M. and S.H.; methodology, G.M. and
Z.Z.; conceptualization, Y.W.; data curation, N.P. and X.X.; validation, X.X.; project administration,
G.M., Y.W. and S.H.; formal analysis, S.H. and L.G.; resources, C.W. and L.G.; supervision, C.W.;
software, L.G.; writing—review and editing. All authors have read and agreed to the published
version of the manuscript.
Funding: This research was funded by the project supported by the State Grid Hebei Electric Power
Co., Ltd. technology projects (kj2023-021). The funder was involved in the study design, collection,
analysis, interpretation of data, the writing of this article or the decision to submit it for publication.
Data Availability Statement: The original contributions presented in the study are included in the
article, further inquiries can be directed to the corresponding author.
Conflicts of Interest: Authors Guozhen Ma, Ning Pang, Yunjia Wang, Shiyao Hu, Xiaobin Xu, Zeya
Zhang were employed by the company State Grid Hebei Electric Power Co., Ltd. Economic and
Technological Research Institute. The remaining authors declare that the research was conducted
in the absence of any commercial or financial relationships that could be construed as a potential
conflict of interest.
References
1.
The chinese government website. Notice of the National Development and Reform Commission, the National Energy Administra-
tion, the Ministry of Agriculture and Rural Affairs on Organizing and Carrying out the Action of Controlling Wind in Thousands
of Villages and Towns. Available online: https://www.gov.cn/zhengce/zhengceku/202404/content_6943057.htm (accessed on
25 March 2024).
2.
Habib, S.; Kamarposhti, M.A.; Shokouhandeh, H.; Colak, I.; Barhoumi, E.M. Economic dispatch optimization considering
operation cost and environmental constraints using the HBMO method. Energy Rep. 2023,10, 1718–1725. [CrossRef]
3.
Wang, H.; Wu, X.; Sun, K.; Du, X.; He, Y.; Li, K. Economic Dispatch Optimization of a Microgrid with Wind–Photovoltaic-Load-
Storage in Multiple Scenarios. Energies 2023,16, 3955. [CrossRef]
4.
Er, G.; Soykan, G.; Canakoglu, E. Stochastic optimal design of a rural microgrid with hybrid storage system including hydrogen
and electric cars using vehicle-to-grid technology. J. Energy Storage 2024,75, 109747. [CrossRef]

Energies 2024,17, 5429 15 of 16
5.
Sharifian, Y.; Abdi, H. Solving multi-area economic dispatch problem using hybrid exchange market algorithm with grasshopper
optimization algorithm. Energy 2023,267, 126550. [CrossRef]
6.
Wang, X.; Wang, S.; Ren, J.; Song, Z.; Zhang, S.; Feng, H. Optimizing Economic Dispatch for Microgrid Clusters Using Improved
Grey Wolf Optimization. Electronics 2024,13, 3139. [CrossRef]
7.
Singh, N.; Chakrabarti, T.; Chakrabarti, P.; Margala, M.; Gupta, A.; Praveen, S.P.; Krishnan, S.B.; Unhelkar, B. Novel Heuristic
Optimization Technique to Solve Economic Load Dispatch and Economic Emission Load Dispatch Problems. Electronics 2023,12,
2921. [CrossRef]
8.
Can, C.; Xiao, F.; Wen, Z.; Yin, S.; Peng, Z.; Xiao, W.; Yuan, M.; Jun, W. Two-staged Generation-grid-load-energy Storage
Interactive Optimization Operation Strategy for Promotion of Distributed Photovoltaic Consumption. Power Syst. Technol. 2022,
46, 3786–3799.
9.
Yang, Y.; Yang, P.; Zhao, Z.; Lai, L.L. A Multi-Timescale Coordinated Optimization Framework for Economic Dispatch of
Micro-Energy Grid Considering Prediction Error. IEEE Trans. Power Syst. 2024,39, 3211–3226. [CrossRef]
10.
Fu, X.; Zhou, Y. Collaborative Optimization of PV Greenhouses and Clean Energy Systems in Rural Areas. IEEE Transactions on
Sustain. Energy 2023,14, 642–656. [CrossRef]
11.
Liu, Z.; Su, M.; Sun, Y. Optimal criterion and global/sub-optimal control schemes of decentralized economical dispatch for AC
microgrid. Int. J. Electr. Power Energy Syst. 2019,104, 38–42. [CrossRef]
12.
Hossain, S.J.; Paul, T.G.; Bisht, R. An Integrated Battery Optimal Power Dispatch Architecture for End-User-Driven Microgrid in
Islanded and Grid-Connected Mode of Operation. IEEE Trans. Ind. Appl. 2018,54, 3806–3819. [CrossRef]
13.
Li, Z.; Xiong, J. Reactive Power Optimization in Distribution Networks of New Power Systems Based on Multi-Objective Particle
Swarm Optimization. Energies. 2024,17, 2316. [CrossRef]
14.
Dong, L.; Lin, H.; Qiao, J.; Zhang, T.; Zhang, S.; Pu, T. A coordinated active and reactive power optimization approach for
multi-microgrids connected to distribution networks with multi-actor-attention-critic deep reinforcement learning. Appl. Energy
2024,373, 123870. [CrossRef]
15.
Nakabi, T.A.; Toivanen, P. Deep reinforcement learning for energy management in a microgrid with flexible demand. Sustain.
Energy Grids Netw. 2021,25, 100413. [CrossRef]
16.
Tang, Z.; Hill, D.J.; Liu, T. Fast Distributed Reactive Power Control for Voltage Regulation in Distribution Networks. IEEE Trans.
Power Syst. 2019,34, 802–805. [CrossRef]
17.
Danalakshmi, D.; Gopi, R.; Hariharasudan, A. Reactive Power Optimization and Price Management in Microgrid Enabled with
Blockchain. Energies 2020,13, 6179. [CrossRef]
18.
Chen, S.; Hu, W.; Du, Y. Three-stage relaxation-weightsum-correction based probabilistic reactive power optimization in the
distribution network with multiple wind generators. Int. J. Electr. Power Energy Syst. 2022,141, 108146. [CrossRef]
19.
Bolognani, S.; Zampieri, S. A Distributed Control Strategy for Reactive Power Compensation in Smart Microgrids. IEEE Trans.
Autom. Control 2013,58, 2818–2833. [CrossRef]
20.
Ma, X.; Liang, C.; Dong, X.; Li, Y. Multi-objective reactive power optimization strategy of power system considering large-scale
renewable integration. Front. Energy Res. 2023,11, 1213154. [CrossRef]
21.
Liu, W.; Long, W.; Gong, Q. Optimal Configuration Method of Dynamic and Static Reactive Power Equipment for Distribution
Network Considering the Load Characteristic of Microgrid. Power Capacit. React. Power Compens. 2019,40, 63–70.
22.
Shi, W.; Lv, X.; He, Y. Distributed Event-Triggered Optimal Algorithm Designs for Economic Dispatching of DC Microgrid with
Conventional and Renewable Generators: Actuator-Based Control and Optimization. Actuators 2024,13, 290. [CrossRef]
23.
Lu, Z.P.; Luo, A.; Zhou, K. Combined Application of DSTATCOM and Microgrid in Reactive Power and Voltage Collaborative
Control on Distribution Network. Proc. CSEE 2010,30, 18–24.
24.
Xu, Y.; Yan, X.; Li, R. Joint Optimization of Active and Reactive Powers in Active Distribution Network With “Generation-grid-
load-energy storage” Interaction in Power Market Environment. Power Syst. Technol. 2019,43, 3778–3789.
25.
Gao, Y.; Ai, Q. Demand-side Response Strategy of Multi-microgrids Based on an Improved Co-evolution Algorithm. Csee J. Power
Energy Syst. 2021,7, 903–910. [CrossRef]
26.
Li, X.; Zhao, W.; Lu, Z. Hierarchical Optimal Reactive Power Dispatch for Active Distribution Network with Multi-microgrids. J.
Electr. Eng. Technol. 2023,18, 1705–1718. [CrossRef]
27.
Liang, Y.; Zhi, L.; Meng, W. Bi-level programming optimal scheduling of ADN with a multi-microgrid based on optimal power
flow. Power Syst. Prot. Control 2020,48, 27–37.
28.
Fan, W.; Huang, L.; Tan, Z.; Xue, F.; De, G.; Song, X.; Cong, B. Multi-objective Optimal Model of Rural Multi-energy Complemen-
tary System with Biogas Cogeneration and Electric Vehicle Considering Carbon Emission and Satisfaction. Sustain. Cities Soc.
2021,74, 103225. [CrossRef]
29.
Jin, X.; Wang, X.; Mu, Y.; Jia, H.; Xu, X.; Qi, Y. Optimal Scheduling Method for a Combined Cooling, Heating and Power Building
Microgrid Considering Virtual Storage System at Demand Side. Proc. CSEE 2017,37, 581–591.
30.
Meng, X.; Gao, J.; Sheng, W.; Gu, W.; Fan, W. Day-Ahead Two-Stage Optimal Scheduling Model for Distribution Network
Containing Distributed Generations. Power Syst. Technol. 2015,39, 1294–1300.
31.
Ming, Z. Protected AgricultureEffects of Light on Growth and Development of Strawberry in Greenhouse; Shihezi University: Shihezi,
China, 2019.

Energies 2024,17, 5429 16 of 16
32.
Jing, Z.; Huan, L.; Yan, S.; Xue, W.; Lei, H.; Mei, L. Effects of Carbon-rich on Quality and Growth Characteristics of Fruit in
Strawberry. J. Shanxi Agric. Sci. 2018,46, 538–543.
33.
He, X.; Hao, R.; Huang, Y.; Jing, T.; Zhou, L. United Optimal Dispatching Control of Photovoltaic and Multi-energy Storage for
Facility Agriculture. Agric. Eng. 2020,10, 49–53.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.