https://doi.org/10.31449/inf.v48i12.6005 Informatica 48 (2024) 171 –184 171 
Reservoir Irrigation Operation Design Based on Dijkstra Algorithm 
Combined with ACO Algorithm 
 
Baoping Jia 
Engineering Management Department, Shanxi Conservancy Technical Institute, Taiyuan 030027, China 
E-mail: jbp1027@163.com 
Keywords: dijkstra algorithm, aco algorithm, reservoir irrigation, path planning, operation model 
Received: April 10, 2024 
In response to the problems of low resource utilization efficiency and imprecise management in 
traditional reservoir irrigation operation systems, a reservoir irrigation operation system combining 
Dijkstra algorithm and ant colony optimization algorithm is proposed. In the study, the improved 
Dijkstra algorithm is first used to optimize the irrigation operation path of the reservoir. On this basis, 
a more comprehensive reservoir irrigation operation system model is constructed by further combining 
ant colony optimization algorithm and improved Dijkstra algorithm. The results showed that the data 
processing accuracy, recall rate, and F1 value of the irrigation operation system model are 95.06%, 
89.68%, and 85.07%, respectively. The length of the optimal path is 21.93 meters, and the annual 
average irrigation water consumption is 131 million m3. The data processing ability and irrigation 
water consumption prediction ability are better than the comparison method. This indicates that 
through reasonable operation design, the system model can better balance the irrigation, power 
generation, and various other needs of the reservoir, improve the efficiency of water resource 
utilization, and promote the sustainable development of agriculture and energy. 
Povzetek: Opisan je inovativni model za načrtovanje namakanja z uporabo rezervoarjev, ki združuje 
Dijkstrov algoritem in algoritem optimizacije s kolonijo mravelj. S tem prispeva k trajnostnemu 
razvoju kmetijstva in energetike.
1 Introduction 
With the development of social economy and population 
growth, water resources have increasingly become an 
indispensable resource in human production and life. As 
an important facility for regulating water resources, the 
reservoir irrigation operation design is of great 
significance for improving water resource utilization 
efficiency, ensuring agricultural production and 
ecological balance. However, reservoir irrigation 
operation is a complex problem that requires 
consideration of various factors, such as reservoir storage, 
discharge, irrigation demand, power generation demand, 
etc. Traditional operation methods are often based on 
experience or simple mathematical models, making it 
difficult to cope with complex and ever-changing 
reservoir operating environments [1-3]. Therefore, 
studying better methods for reservoir irrigation operation 
has important practical significance and theoretical value. 
The Dijkstra algorithm is a classic shortest path algorithm 
mainly used to solve the shortest path problem in graphs 
with non negative weights [4-5]. Ant Colony 
Optimization (ACO) is an algorithm that simulates ants 
searching for food paths, using information exchange 
between ants to find the optimal path. The combination of 
Dijkstra algorithm and ACO algorithm can effectively 
solve the irrigation operation problem of reservoirs [6-7]. 
The study first utilizes the improved Dijkstra algorithm to 
optimize the irrigation operation path of the reservoir, and 
then combines the ACO algorithm and the improved 
Dijkstra algorithm to construct a model of the reservoir 
irrigation operation system based on the path 
optimization. The research aims to better address 
multi-objective and multi-constraint problems in reservoir 
irrigation operation, and provide better solutions for 
practical applications. 
The first part of the study introduces the current 
research status of reservoir irrigation operation and 
machine learning in reservoir operation. The second part 
of the study utilizes the improved Dijkstra algorithm and 
ACO algorithm to construct a reservoir irrigation 
operation model. The third part verifies the performance 
of the constructed reservoir irrigation operation model 
through simulation experiments. The fourth part 
summarizes the experimental results and analyzes the 
advantages and disadvantages of the research methods 
used. 
The design of reservoir irrigation operation is an 
important issue in the field of agricultural water 
conservancy, with the goal of rational allocation of water 
resources, meeting irrigation needs while maximizing 
water resource utilization efficiency. With the 
development of technology, more and more intelligent 
algorithms are being applied to the design of reservoir 
irrigation operation to solve its multi-objective and 

172   Informatica 48 (2024) 171 –184                                                                     B. Jia 
multi-constraint optimization problems. Therefore, 
numerous experts and scholars have conducted extensive 
research on reservoir irrigation operation. Bakanos and 
Katsifarakis conducted applied research on the urban 
water supply, irrigation, hydropower generation, flood 
control, and environmental flow protection of reservoirs 
based on the study of a multi-purpose reservoir operation 
system. The system simulated three different hydrological 
scenarios by optimizing genetic algorithms. The results 
showed that the system was able to successfully identify 
and quantify the increase in hydropower production, and 
pointed out the necessary changes in the reservoir 
operation rule curve [8]. Boluwade conducted water 
resource operation research using global precipitation 
measurement tasks and a comprehensive multi satellite 
inversion system to effectively estimate and schedule 
precipitation. The use of CHIRPS and TAMSAT systems 
was studied for seasonal forecasting, irrigation operation, 
and reservoir operation analysis. The results indicated 
that the system could provide more decision-making 
reference data for water resource managers and 
decision-makers [9]. Waluyadi et al. optimized the 
operation model of reservoirs by using multiple linear 
regression and stochastic models. A distribution 
probability modeling was conducted on the operation of 
inflow, irrigation, and raw water supply using a model. 
The results indicated that the reliability and elasticity of 
the model in reservoir operation had the highest benefits 
and performance [10]. 
Chek proposed a linear regression method based on 
Dijkstra algorithm to find the shortest path from the 
source node to the target node. This method could reduce 
computation time and memory consumption, and 
achieves decentralized task allocation path planning. The 
results indicated that this method could effectively avoid 
collisions and prevent the possibility of deadlock, and 
provide a better AGV route [11]. Li et al. proposed an 
exploration path planning method based on information 
entropy to improve the exploration efficiency and map 
creation performance of unknown scenes. The study first 
constructed an information entropy map and set 
boundaries, then used Dijkstra's algorithm to explore 
paths starting from boundary features, and finally drew a 
map based on the results of path exploration. The results 
indicated that the path planning algorithm constructed in 
the study not only achieved high detection efficiency, but 
also had an increased coverage range [12]. Wang et al. 
proposed a time sensitive network operation algorithm 
based on ACO algorithm to improve the timeliness and 
accuracy of operation in the industrial role. Research 
focused on end-to-end delay manipulation of time 
triggered flows in time sensitive networks to address 
errors in time triggered flows. The results indicated that 
the algorithm could significantly improve convergence 
speed and avoid the occurrence of local optimal solutions 
[13]. The summary of the relevant research conducted by 
the above-mentioned experts and scholars is shown in 
Table 1. 
 
Table 1: Summary of research results by relevant experts and scholars 
Literature Proposed models/methods Advantage Shortcoming 
Bakanos and Katsifarakis 
[8] 
A multipurpose reservoir 
dispatching system based on 
genetic algorithm optimization 
Suitable for 
multi-purpose reservoirs 
High computational 
complexity 
Boluwade [9] 
A water resource scheduling 
model using global precipitation 
measurement tasks and 
integrated multi satellite 
inversion systems 
Provided more 
decision-making 
reference data for water 
resource managers and 
decision-makers 
Accuracy and coverage 
are limited 
Waluyadi et al. [10] 
Reservoir operation 
optimization model based on 
multiple linear regression and 
stochastic model 
High reliability and 
elasticity demonstrated 
in reservoir scheduling 
High model complexity 
Chek [11] 
Linear regression method based 
on Dijkstra algorithm for path 
planning 
Plan task allocation and 
paths in reservoir 
scheduling 
Possible inability to 
capture non-linear 
relationships 
Li et al. [12] 
Exploration of reservoir 
operation path planning method 
based on information entropy 
Can directly optimize 
scheduling strategies 
The calculation of 
information entropy 
involves a large amount 
of data processing 
Wang et al. [13] 
A time sensitive network 
scheduling algorithm based on 
aco algorithm 
Can solve the errors in 
time triggered flow 
The ACO algorithm is 
affected by parameter 
settings 
 
In summary, it is of great significance to study the optimization of reservoir irrigation scheduling systems 

Reservoir Irrigation Operation Design Based on Dijkstra Algorithm …                Informatica 48 (2024) 171 –184  173 
using Dijkstra algorithm and ACO algorithm. The study 
of integrating two algorithms can improve the reservoir 
irrigation scheduling system's ability to handle 
multi-objective and multi constraint scheduling problems. 
Meanwhile, it also can avoid the increased path planning 
time caused by the processing process, providing better 
solutions for practical applications. 
 
2 Design of reservoir irrigation 
operation system combining 
dijkstra algorithm and ACO 
algorithm 
The reservoir irrigation operation system is a complex 
system engineering that involves multiple issues such as 
optimal allocation of water resources and ecological 
environment protection. To achieve efficient and 
environmentally friendly reservoir irrigation operation, a 
new reservoir irrigation operation system model is 
constructed by combining Dijkstra algorithm with ACO 
algorithm. This system can dynamically adjust irrigation 
paths and resource allocation according to actual 
situations, improving irrigation efficiency and water 
resource utilization. 
 
2.1 Reservoir irrigation operation path based 
on improved dijkstra algorithm 
To ensure the sustainability and scientificity of reservoir 
irrigation operation, the process of reservoir resource 
operation is carried out on a periodic basis. Reservoir 
resource operation is a complex process involving 
multiple factors, including water demand, water quality 
conditions, climate factors, ecological balance, economic 
factors, regulatory policies, and social impacts. These 
factors need to be comprehensively considered, 
scientifically analyzed, and reasonable decisions made to 
achieve the rational allocation and sustainable utilization 
of reservoir resources. The main parameter variables 
involved in the process of reservoir resource operation 
include reservoir status, decision quantity, etc. [14-15]. 
Irrigation demand reflects the water demand of crops and 
is an important basis for irrigation operation. Decision 
variables are controllable factors in reservoir 
management, such as the outflow and flood discharge of 
the reservoir. When operate reservoir resources, it is 
necessary to determine the state variables of the reservoir, 
which can be represented by formula (1). 
 
1 t t t t
V V Q q
+
= + −        (1) 
 
In formula (1), 
1 t
V
+
 represents the state variable of 
the reservoir. 
t
V represents the water volume of the 
reservoir before the operation of the reservoir resources at 
moment t . 
t
Q represents the amount of water entering 
the reservoir at the moment t after the operation 
decision. 
t
q represents the discharge volume of the 
reservoir at moment t . Through the analysis of reservoir 
resources, it is found that in the process of irrigation 
operation, it is necessary to evaluate the benefits of 
irrigation operation, evaluate the actual effectiveness of 
irrigation operation schemes, compare the advantages and 
disadvantages of different schemes, and provide scientific 
basis for decision-makers. To effectively evaluate the 
operation efficiency, constraints are applied to both flood 
season and non-flood season, and the optimal objective 
function is obtained by combining recursive equations, as 
shown in formula (2). 
 
1
( ) max( ( , , ) ( ))
t t t t t t t
E V f V Q q E V
+
=+ (2) 
 
In formula (2), () f  represents the functional 
relationship. 
t
E represents the optimization strategy at 
t time. After completing the study of reservoir operation 
coefficients and objective functions, the selection of 
irrigation paths also has a direct impact on irrigation. This 
is because the location of the reservoir has the most direct 
impact on irrigation costs, so the study applies the 
Dijkstra algorithm to the path optimization of reservoir 
irrigation operation. The basic idea of the Dijkstra 
algorithm is to start from the starting point, gradually 
expand outward to adjacent nodes, and select the edge 
with the shortest distance to expand the node until it 
reaches the target node. The main feature of the algorithm 
is the use of greedy strategy, which selects the current 
shortest distance each time and updates the distance of all 
adjacent nodes. 
In the application of Dijkstra algorithm, only the 
distance between nodes is usually considered. However, 
in actual path planning, it is more important to consider 
the time and cost required for transportation between 
different nodes. Therefore, the study conducts an in-depth 
analysis of the traditional Dijkstra algorithm and 
improves it to better apply it to the optimization problem 
of reservoir irrigation paths [16-17]. When using 
reservoir water resources for actual irrigation operation, 
the location of the reservoir is the main factor directly 
affecting operation. The study comprehensively considers 
labor costs, losses, etc., and improves the Dijkstra 
algorithm. The flowchart for improving the Dijkstra 
algorithm is shown in Figure 1. 

174   Informatica 48 (2024) 171 –184                                                                     B. Jia 
End the 
algorithm 
and return to 
the optimal 
scheduling 
plan
Labor costs
Select nodes 
and classify 
them
Loss expenses
Determine the 
weight of 
comprehensive 
expenses
Update scheduling 
information from 
the source reservoir 
node to other nodes
Merge update sets 
for classification
Does it 
include 
all nodes
 
Figure 1: Flowchart for improving Dijkstra algorithm 
 
The study will improve the Dijkstra algorithm and 
apply it to actual reservoir irrigation operation. It is found 
that the time, manpower, and economic costs required for 
irrigation operation from Reservoir A to Reservoir B are 
not exactly the same as those required for operation from 
Reservoir B to Reservoir A. This is mainly because 
operation in two directions may face different 
geographical environments, hydrological conditions, 
traffic conditions, and other factors, all of which may 
have an impact on the time, manpower, and economic 
costs of operation. Based on this, to ensure the 
scientificity of the research, when using the improved 
Dijkstra algorithm for irrigation path planning, the 
average cost between reservoir nodes is taken as the 
standard cost value. At this point, the labor cost value can 
be calculated, which can be represented by formula (3). 
12
1 1 2 2
21
( ) ( )
CC
C C C C C
CC
= + + +
    (3) 
In formula (3), C represents the average labor cost 
value. 
1
C represents the labor cost value of A reservoir. 
2
C represents the labor cost value of Reservoir B. After 
calculating the labor cost value, it is also necessary to 
calculate the final expenses, and the final total loss value 
can be calculated using formula (4). 
12
1 1 1 2 2 2
21
( ( ) ( ))
CC
R S W C C W C C L
CC
= + + + +
(4) 
In formula (4), R represents the total loss value. 
S represents the distance between reservoirs. 
1
W 
represents the water resource loss value of Reservoir A. 
2
W represents the water resource loss value of Reservoir 
B. L represents other loss values during the operation 
process. The comparison results before and after 
optimizing the irrigation operation path are shown in 
Figure 2. 
0
1.0
2.0
4.0
6.0
8.0
Billion m
3
/year
Improved Dijkstra 
algorithm
Water inflow
Yield
7.0
5.0
3.0
Traditional 
Dijkstra algorithm
(a) Comparison results of inflow and 
outflow before and after optimizing 
irrigation scheduling paths
20
30
40
60
80
100
Utilization rate/%
90
70
50
Improved Dijkstra 
algorithm
Traditional 
Dijkstra algorithm
(b) Comparison results of inflow and 
outflow before and after optimizing 
irrigation scheduling paths
 
Figure 2: Comparison results of irrigation operation path optimization before and after 

Reservoir Irrigation Operation Design Based on Dijkstra Algorithm …                Informatica 48 (2024) 171 –184  175 
 
From Figure 2, after optimizing the irrigation path, 
the utilization of inflow and outflow of water resources in 
the reservoir has been significantly improved, and the 
utilization rate of water has also been significantly 
improved. This indicates that the optimization of 
reservoir irrigation operation paths based on the improved 
Dijkstra algorithm has a positive effect. 
 
2.2 Improved Dijkstra algorithm combined 
with ACO algorithm for irrigation operation 
system model construction 
Through the study of using the improved Dijkstra 
algorithm for reservoir irrigation operation, it is found 
that the Dijkstra algorithm mainly solves the single 
objective optimization problem, that is, finding the 
shortest path. However, there may be multiple conflicting 
objectives in irrigation operation, such as irrigation area, 
water consumption, energy consumption, etc. [18-19]. To 
address these issues, the ACO algorithm is introduced in 
the study. In the irrigation operation problem of 
reservoirs, the irrigation channel network can be 
abstracted as a graph model, where nodes represent 
reservoirs, pumping stations, irrigation areas, etc. The 
edge represents the water flow path, and the weight of the 
edge can represent the time, cost, or energy consumption 
required for the water flow to pass through this path. 
Then, the ACO algorithm is used to search for the 
optimal irrigation path from the starting reservoir to the 
target reservoir on the graph model. The flowchart of the 
ACO algorithm is shown in Figure 3. 
 
Does the 
loop end
Whether to 
terminate
Set initialization 
parameters and 
constraints
Ant selection path
Partial updates
Identify the optimal 
solution
Global updates
Y
N
Y
N
 
Figure 3: ACO algorithm flowchart 
 
 
When integrating the ACO algorithm with the 
improved Dijkstra algorithm, the first step is to set the 
objective function of the operation system model, which 
can be represented by formula (5). 
2
,
min ( ) ( )
ij
F R t D t  =−

        (5) 
In formula (5), 
,
()
ij
Rt
 represents the expected 
probability value of the i th and 
j
th nodes in the t 
period. () Dt represents the actual probability value 
during the t period. At this point, the moderate function 
can be calculated using formula (6). 
22
11 max max
( ) ( ) ( ) ( )
( ) ( )
k
kk
ij
Fit
R t D t R t D t
D i D j
==
=
    −−
   
+

 (6) 
 
 
In formula (6), t represents the total amount of 
information obtained by k ants in the ant colony during 
the t period. 
max
() Di represents the total amount of 
information on the path where node i is located. 
max
() Dj represents the total amount of information on 
the path where node 
j
 is located. Optimizing the 
objective function and moderate function can avoid the 
slow initial speed of the ACO algorithm, and state 
transition rules need to be considered in the process of 
using the ACO algorithm for path planning. To regulate 
the concentration of ants in searching paths, a harmonic 
parameter is added to the search rules to reduce the 
probability of ants deviating from the concentrated area 
and selecting paths. The transfer rule formula obtained by 
adding harmonic parameters is shown in formula (7). 
 

176   Informatica 48 (2024) 171 –184                                                                     B. Jia 
,
( ) ( )
()
( ) ( )
ab
ij ij
ij ab
ij ij
G t B t
p t s
G t B t
=

       (7) 
In formula (7), 
s
 represents a harmonic parameter 
with a range of values between [0,1] and a random 
number. 
,
()
ij
pt
 represents the probability value of state 
transition. 
()
ij
Gt
 represents the total amount of 
pheromones accumulated on paths i and 
j
 during the 
t period. 
()
ij
Bt
 represents the estimated moderate 
function value in the t period based on the objective 
function. 
a
 and b represent the parameters of 
different nodes, respectively. By applying new state 
transition rules, the probability of ants exploring 
redundant paths can be reduced, thereby improving the 
exploration efficiency of the algorithm and saving time. 
The flow chart of irrigation operation path optimization 
based on improved Dijkstra algorithm and ACO 
algorithm is shown in Figure 4. 
Start
Initialize fitness 
function, 
number of 
iterations, and 
other parameters
Randomly 
generate 
feasible 
solution 
populations
Calculate 
population 
fitness
i ＜cMax
Choose 
the best 
individual
i＜cMax
Evaluate the 
population 
and choose 
the optimal 
path
Reliability 
factor
Volatility 
coefficient
Y
N
Y
N
Y
N
Determine 
the next 
node 
update
Ants at 
nodes
Recalculate 
Access
To access 
nodes
Initialize 
parameter 
ant colony 
pheromones 
separately
ACO 
optimization
 
Figure 4: Irrigation operation path optimization flowchart based on improved Dijkstra algorithm and ACO algorithm 
 
Through the optimization of irrigation operation 
paths, it is found that when using the ACO algorithm to 
optimize the path, the volatilization of pheromones is 
reduced, which can affect the accuracy of ants in path 
planning. Therefore, the study introduces a volatilization 
factor and improves the local and new rules to accelerate 
the convergence speed of the algorithm, thereby avoiding 
ants appearing on the same path during the convergence 
process. The improved local update rule can be 
represented by formula (8). 
( 1) (1 ) ( ) ( , 1)
ij ij
G t dis G t dis t t +  − +  +
 (8) 
In formula (8), 
( 1)
ij
Gt +
 represents the total 
amount of pheromones accumulated on paths i and 
j
 
during the 1 t + period. dis represents the volatility 
factor, which is a random number between [0,1]. 
( , 1) dis t t + represents the increase in pheromones at t 
and 1 t + moments after introducing volatilization 
factors. After completing the local update, it is necessary 
to update the global path, which can improve the 
guidance weight of pheromones. By introducing global 
update coefficients, the concentration of pheromones can 
be increased, significantly improving the running speed 
of the algorithm. The global update rule can be 
represented by formula (9). 
, ,
1
( , )
( , 1)
0
ij ij
il
ad L G t t


  + =



     (9) 
In formula (9), ad represents the global update 
coefficient, which is a random number between [0,1]. 
, ij
L
 represents the length of the global path. After 
optimizing the irrigation operation path of the reservoir, 
the construction time and engineering standards of 
different reservoirs are different in the construction of the 
water resources network. The study introduces reliability 
parameters for coordination, which are comprehensively 
set based on the construction time and intensity of the 
reservoir. The longer the construction time, the worse the 
reliability. It is assumed that the lower the intensity, the 
worse the reliability [20-21]. Therefore, the study 
introduces a reliability coefficient in the selection of 
reservoir nodes to reduce the irrigation operation tasks of 

Reservoir Irrigation Operation Design Based on Dijkstra Algorithm …                Informatica 48 (2024) 171 –184  177 
reservoirs with aging facilities and low storage capacity. 
The calculation formula for reliability coefficient can be 
represented by formula (10). 
D
HV
T
=           (10) 
In formula (10), H represents the reliability 
coefficient value. After obtaining the reliability 
parameters, a new irrigation operation probability 
formula can be obtained, as shown in formula (11). 
,
( ) ( )
()
( ) ( )
ab
ij ij
ij ab
ij ij
G t B t
p t H s
G t B t
=

    (11) 
Based on the above research, it is necessary to 
clarify the design principles before constructing a water 
resources operation system model. Firstly, it needs to 
meet the downstream urban living, industrial and 
agricultural water use, and urban ecological needs. 
Secondly, it needs to consider the water level difference 
between flood season and non flood season. Finally, 
hydrological data is stored and analyzed simultaneously 
for decision-making purposes [22]. After requirement 
analysis, the simulation system functions are divided into 
three parts: conventional operation, optimized operation, 
and manual operation. The flowchart of the irrigation 
operation system model combining the improved Dijkstra 
algorithm with the ACO algorithm is shown in Figure 5. 
Irrigation 
demand
Supply capacity 
assessment
Irrigation path 
planning
Water supply 
demand data
Input 
hydrological data
Input 
reservoir 
network 
data
Water scheduling 
model
Output irrigation 
scheduling plan
Flood control plan
Power 
generation plan
 
Figure 5: Flow chart of an improved irrigation operation system model combining Dijkstra algorithm with ACO 
algorithm 
 
3 Performance analysis of irrigation 
operation system model based on 
Dijkstra and ACO 
To verify the performance of the reservoir irrigation 
operation system model constructed based on Dijkstra 
algorithm and ACO algorithm, this study compared the 
Greedy Algorithm (GA), Particle Swarm Optimization 
(PSO), and Dijkstra+ACO algorithms. The accuracy, 
recall, F1 value, optimal path length, number of search 
nodes, and actual irrigation pre measurement in path 
planning data processing were used as validation 
indicators for performance verification. 
3.1 Path and irrigation demand performance 
analysis of reservoir irrigation operation 
system model 
To verify the path planning performance of the irrigation 
operation system model, simulation experiments were 
conducted using actual data from a certain reservoir. The 
reservoir had a regulating capacity of 780 million m3, an 
average annual runoff of 1.15 billion m3, a normal 
storage level of 85 meters, and a dead water level of 65 
meters. The installed power generation capacity of the 
reservoir was 40MW, and the designed irrigation area 
was 65000hm2. The power generation guarantee rate was 
designed to be 85%, and the limited water supply during 

178   Informatica 48 (2024) 171 –184                                                                     B. Jia 
the designed irrigation failure period was reduced to 60% 
of the normal water supply. Simultaneously comparing 
GA, PSO, and Dijkstra+ACO algorithms could better 
evaluate the performance and effectiveness of the model. 
The comparison results of accuracy, recall, and F1 values 
of the three algorithms are shown in Figure 6. 
 
0.6
0.7
0.8
0.9
1.0
GA
Accuracy/%
Number of 
experiments
PSO
100
200
300
400
Different methods 
Dijkstra
+ACO
0.6
0.7
0.8
0.9
1.0
100
200
300
400
0.6
0.7
0.8
0.9
1.0
100
200
300
400
GA
Recall rate/%
Number of 
experiments
PSO
Different methods 
Dijkstra
+ACO
GA
F1 value
Number of 
experiments
PSO
Different methods 
Dijkstra
+ACO
 
Figure 6: Comparison results of accuracy, recall, and F1 values of three algorithms 
 
According to Figure 6, the data processing accuracy, 
recall, and F1 values of Dijkstra+ACO algorithm, PSO 
algorithm, and GA in irrigation operation path planning 
process were 95.06%, 89.68%, 85.07%, 96.27%, 90.16%, 
88.29%, and 0.92, 0.86, and 0.83, respectively. The 
comparison showed that the irrigation operation system 
model constructed using Dijkstra+ACO algorithm had 
higher robustness and stability in the processing of 
irrigation path data. To analyze the temporal and spatial 
complexity of Dijkstra+ACO algorithm in reservoir 
scheduling, the study took the execution time and the size 
of memory required in the irrigation path planning 
process as the indicators of temporal and spatial 
complexity analysis. The results of the comparison of 
execution time and required memory of the three 
algorithms are shown in Table 2. 
 
Table 2: Comparison results of execution time and memory required for three algorithms 
Arithmetic Execution time/s Memory size required/MB 
GA 120.59 150.5 
PSO 113.18 120.7 
Dijkstra+ACO 92.63 108.5 
 
As analyzed in Table 2, in the time complexity 
comparison, the execution time of the three algorithms 
had more obvious differences. The execution time of the 
Dijkstra+ACO algorithm was shorter than that of the GA 
and PSO algorithms by 20.55s and 27.96s, respectively. 
In the memory consumption of the path planning, the 
Dijkstra+ACO algorithm consumed 12.2MB and 42MB 
less memory than the GA and PSO algorithms, 
respectively. This indicated that the Dijkstra+ACO 
algorithm constructed in the study was able to find the 
optimal or near-optimal irrigation path planning scheme 
in a shorter period of time and complete the 
computational task with less memory consumption. To 
analyze the uncertainty and sensitivity of Dijkstra+ACO 
algorithm in reservoir scheduling, the ability to analyze 
unexpected irrigation demand (out of 10 points) and 
irrigation elasticity coefficient were used as validation 
indexes. The results of uncertainty and sensitivity 

Reservoir Irrigation Operation Design Based on Dijkstra Algorithm …                Informatica 48 (2024) 171 –184  179 
analysis of the three algorithms are shown in Table 3. 
 
Table 3: Uncertainty and sensitivity analysis results of the three algorithms 
Arithmetic 
Capacity to analyze unforeseen 
irrigation needs/points 
Irrigation elasticity coefficient 
GA 7.5 0.79 
PSO 8.2 0.81 
Dijkstra+ACO 8.9 0.85 
 
From the analysis in Table 3, in the comparison of 
accidental irrigation demand analysis capability, the 
Dijkstra+ACO algorithm scored 8.9 for accidental 
irrigation demand analysis capability, while the PSO and 
GA scores were 8.2 and 7.5, respectively. The irrigation 
elasticity coefficients of the three algorithms of 
Dijkstra+ACO algorithm, PSO and GA had the values of 
0.85, 0.81, and 0.79, respectively. This further validated 
the effectiveness and superiority of Dijkstra+ACO 
algorithm in the field of reservoir scheduling and 
provided a more efficient, stable, and flexible solution for 
reservoir scheduling. To further verify the performance of 
the irrigation operation system model in irrigation path 
planning, the optimal path length and path planning were 
studied as validation indicators. The comparison results 
of the optimal path length and path planning for three 
methods are shown in Figure 7. 
GA
PSO
Dijkstra+ACO
0 100 90 80 70 60 50 40 30 20 10
24.0
Iterations
(a) Comparison results of the optimal 
path of three methods
23.8
23.6
23.4
23.2
23.0
22.8
22.6
22.4
22.2
22.0
21.8
21.6
Optimal path length/m
GA
PSO
Dijkstra+ACO
0 10 9 8 7 6 5 4 3 2 1
Horizontal path
(b) Comparison results of three optimal 
path planning methods
10
9
8
7
6
5
4
3
2
1
0
Vertical path
 
Figure 7: Comparison results of optimal path length and path planning using three methods 
 
As shown in Figure 7 (a), during the irrigation 
operation, the Dijkstra+ACO algorithm completed the 
search and planning of the optimal path length in 32 
iterations. The PSO algorithm and GA completed the 
optimal path length and planning when iterating 41 and 
57 times, respectively. The optimal path lengths for 
Dijkstra+ACO algorithm, PSO algorithm, and GA were 
21.93m, 22.15m, and 23.6m, respectively. As shown in 
Figure 7 (b), in the comparison results of optimal path 
planning, the number of path nodes in the Dijkstra+ACO 
algorithm was significantly less than that in the PSO 
algorithm and GA, indicating that the Dijkstra+ACO 
algorithm had better search ability and robustness in 
solving path planning problems. To verify the 
performance of the irrigation system model in irrigation 
operation requirements, the sensitivity analysis of 
reservoir capacity and precipitation on irrigation water 
consumption was studied as validation indicators for 
performance testing. The sensitivity analysis comparison 
results of three methods affected by reservoir capacity 
and precipitation are shown in Figure 8. 

180   Informatica 48 (2024) 171 –184                                                                     B. Jia 
12 11 9 6 5 4 3 2 1 0 7 8 10
0
0.5
1.0
1.5
2.0
2.5
Month
Irrigation water 
consumption/billion m
3
(a) Three methods for assessing reservoir 
capacitySensitivity analysis of impact
12 11 9 6 5 4 3 2 1 0 7 8 10
1
2
3
4
5
Month
Irrigation water 
consumption/billion m
3
(b) Three methods for measuring 
precipitationSensitivity analysis of impact
0
Actual value
Dijkstra+ACO
PSO
GA
Actual value
Dijkstra+ACO
PSO
GA
 
Figure 8: Comparison of sensitivity analysis results of three methods affected by reservoir capacity and precipitation 
 
As shown in Figure 8 (a), the actual irrigation water 
consumption also showed a significant change with the 
change of reservoir capacity, with an average annual 
irrigation water consumption of 126 million m3. The 
Dijkstra+ACO algorithm had the smallest difference with 
the actual irrigation value, with an average annual 
irrigation water consumption of 131 million m3. The 
average annual irrigation water consumption of PSO 
algorithm and GA was 137 million m3 and 141 million 
m3, respectively. From Figure 8 (b), when there was a 
change in precipitation, there would also be a significant 
change in irrigation water consumption. The curve 
showed significant fluctuations within the range of 
rainfall variation, indicating a high level of uncertainty in 
rainfall. With the increase of rainfall, the irrigation water 
consumption would also correspondingly increase to 
meet the water requirements of farmland. To further 
verify the performance of the irrigation system model in 
irrigation operation requirements, the study compared the 
actual irrigation demand with the irrigation demand 
predicted by three methods. The comparison results 
between the irrigation demand of three methods and the 
actual demand are shown in Figure 9. 
10
15
20
30
40
50
Irrigation water volume/10
6
m
3
Requirement
Dijkstra+ACO
45
35
25
PSO
GA
1 2 3 4 5 6 7 8 9 10 11 12
Month
 
Figure 9: Comparison of irrigation demand and actual demand using three methods 
 
As shown in Figure 9, there were significant 
differences in the demand for irrigation in different 
months of the year, which was directly related to rainfall 
and crop cultivation. Compared with the actual irrigation 
demand, the demand of Dijkstra+ACO algorithm could 
basically meet the irrigation demand. However, there was 
a certain gap between the irrigation demand of PSO 
algorithm and GA and the actual irrigation demand, 
which could not effectively meet the actual irrigation 
demand. This indicated that the study of using 
Dijkstra+ACO algorithm to construct a reservoir 
irrigation operation system model could meet real needs 
in irrigation path optimization and irrigation demand, and 
had higher practicality. 
 
 
 
 

Reservoir Irrigation Operation Design Based on Dijkstra Algorithm …                Informatica 48 (2024) 171 –184  181 
3.2 Application performance analysis of 
reservoir irrigation operation system model 
To further verify the application performance of the 
reservoir irrigation operation system model, the study 
used the reservoir irrigation operation system model to 
predict the water demand hydrograph and compared it 
with the measured water demand hydrograph. The 
comparison results between the predicted and the actual 
measured water demand hydrographs are shown in Figure 
10. 
 
240 216 192 24 48 72 96 120 144 168 0
0
200
400
600
800
1000
1200
1400
Flow rate (L/s
-1
)
Time/h
Predicted water 
demand hydrograph
Actual measured water 
demand process line
 
Figure 10: Comparison results between predicted and actual measured water demand hydrographs 
 
From Figure 10, in the comparison of the 10-day 
water demand process lines, the average value of the 
actual measured water demand process lines was 
786.3L/s-1, and the interval between the minimum and 
maximum values was [283.7,1195.8] L/s-1. The average 
value of the predicted water demand hydrograph was 
801.6L/s-1, and the interval between the minimum and 
maximum values was [269.7,1089.2] L/s-1. The 
difference between the two was 15.3L/s-1, which had a 
relatively small impact on the practical application of the 
irrigation operation system model. This indicated that the 
reservoir irrigation operation system model constructed in 
the study could also achieve good application results in 
practical applications, and could further meet the needs of 
irrigation operation on the basis of optimizing the 
operation path. To verify the actual predictive 
performance of the irrigation operation system model, the 
study used irrigation restriction water supply line, 
irrigation anti damage line, irrigation comprehensive anti 
damage line, power generation comprehensive anti 
damage line, and anti water discharge line as indicators 
for performance verification. The calculation prediction 
curves for five validation indicators are shown in Figure 
11. 
 
50
55
60
70
80
90
Irrigation water volume/10
6
m
3
85
75
65
1 2 3 4 5 6 7 8 9 10 11 12
Month
0
Irrigation restriction 
water supply line
Irrigation anti damage line
Anti water discharge line
Irrigation comprehensive 
anti damage line
Power generation comprehensive 
anti damage line
 
Figure 11: Calculation and prediction curve of five validation indicators 

182   Informatica 48 (2024) 171 –184                                                                     B. Jia 
 
As shown in Figure 11, the irrigation operation 
system model could accurately predict the irrigation 
restriction water supply line, irrigation anti damage line, 
irrigation comprehensive anti damage line, power 
generation comprehensive anti damage line, and anti 
water discharge line in different months. The effective 
prediction of these five indicators could ensure the 
reasonable allocation of water supply, prevent damage to 
reservoirs, environment, and land, improve water supply 
efficiency, and comprehensively consider various factors 
to achieve sustainable development of irrigation. To 
verify the comprehensive operation performance of the 
reservoir irrigation operation system, the study used 
irrigation and power generation as indicators to verify the 
comprehensive operation performance. The 
comprehensive operation results of the reservoir 
irrigation operation system are shown in Table 4. 
 
 
Table 4: Comprehensive operation results of reservoir irrigation operation system 
Method of 
calculation 
Irrigatio
n water 
supply/b
illion m3 
Annual 
irrigation 
guarante
e rate/% 
Water supply 
rate during 
the 
destruction 
period/% 
Reducing 
water 
supply 
guarantee 
rate/% 
Annual 
average 
power 
generation/(
10000 
kW·h) 
Guarant
eed 
power/(1
0000 
kW·h) 
Guara
nteed 
powe
r/kW 
Power 
generatio
n 
guarante
e rate/% 
Runoff 
regulation 
4.15 85.32 62.37 / 10563 627 719 85 
Comprehensive 
operation of 
irrigation and 
power 
generation 
4.15 85.32 62.37 100 10867 71529 
9256
4 
85 
 
According to Table 4, there was no significant 
difference between the two methods of runoff regulation 
and comprehensive operation of irrigation and power 
generation in terms of irrigation water supply, annual 
irrigation guarantee rate, water supply rate during the 
destruction period, and reducing water supply guarantee 
rate. But in terms of ensuring electricity, the 
comprehensive operation method of irrigation and power 
generation ensured a larger amount of electricity, and at 
the same time, the average annual power generation and 
guaranteed power were also higher. In addition, the 
average annual power generation of the comprehensive 
operation method for irrigation and power generation also 
slightly increased. This indicated that the reservoir 
irrigation operation system could allocate water and 
energy resources more reasonably, improve resource 
utilization efficiency, and achieve optimized resource 
allocation. At the same time, it could also reduce the 
phenomenon of reservoir water discharge and the 
negative impact on the environment. The study, to further 
validate the performance of the irrigation scheduling 
system model, took a reservoir in a hilly area as a case 
study. The reservoir was located in an irrigated 
agricultural area and was also tasked with power 
generation, so its irrigation scheduling design needs to 
take into account the dual demands of irrigation and 
power generation. The study first collected data on 
meteorological data, hydrological data, irrigation demand 
and power generation demand for the past five years for 
this reservoir. Then the irrigation water supply and 
average annual power generation were qualitatively 
evaluated. After the implementation of the new 
scheduling scheme, the irrigation water supply and power 
generation of the reservoir were 10.59 million m3 and 
0.58 million kw, respectively, and the irrigation water 
supply and power generation without the new scheduling 
scheme were 9.96 million m3 and 0.53 million kw, 
respectively. After the use of the new scheduling scheme, 
the irrigation water supply and power generation of the 
reservoir increased by 0.63 million m3 and 0.05 million 
kw, respectively. This indicated that the Dijkstra 
algorithm combined with the ACO algorithm enables the 
model to find the optimal solution among many possible 
scheduling schemes, thus realizing the rational allocation 
and efficient utilization of water resources. In addition, 
the model also considered the dual demand of irrigation 
and power generation, and realized the synergistic 
benefits between them by coordinating the relationship 
between them. 
4 Discussion 
By summarizing and analyzing the above research 
content, this study explored the application of Dijkstra 
algorithm and ACO algorithm in reservoir irrigation 
scheduling design, and successfully constructed an 
irrigation scheduling system model that can handle 
multi-objective and multi constraint optimization 
problems. There are certain differences in the water 
demand process compared to actual cases, which can be 
attributed to multiple aspects. On the one hand, although 
the study combined Dijkstra algorithm and ACO 

Reservoir Irrigation Operation Design Based on Dijkstra Algorithm …                Informatica 48 (2024) 171 –184  183 
algorithm for predicting water demand, changes in 
natural factors such as climate, soil conditions, and crop 
growth status may still have an impact on actual water 
demand, and the model fails to fully capture these 
dynamic changes. On the other hand, the parameter 
settings and algorithm optimization of the model may 
also have an impact on the prediction results. Therefore, 
future research can further explore how to optimize 
model parameters and algorithms to improve prediction 
accuracy. Secondly, the system model has achieved 
significant results in the comprehensive scheduling of 
irrigation and power generation. The system not only 
ensured agricultural irrigation water, but also fully 
considered the demand for power generation, achieving 
effective coordination between the two. However, 
research has also found that the operation and 
management of the system require professional technical 
personnel for operation and maintenance. Therefore, how 
to cultivate and introduce professional talents in related 
fields is one of the key to the successful operation of the 
system. The hybrid algorithm also explored the impacts 
of ecological and social factors, and in adapting to the 
ecological balance, it not only met the needs of 
agricultural production, but also provided the ecosystem 
with the necessary amount of water, which helps to 
maintain the balance of water ecology and biodiversity. 
In adapting to social impacts, by increasing the amount of 
water supplied for irrigation and the annual guaranteed 
rate of irrigation, it ensured stable agricultural production 
and meets the social demand for food and other 
agricultural products. Therefore, the development of the 
agricultural economy can be promoted to meet the 
demand for electricity in industry, commerce and services, 
and promote the overall development of the economy. 
In summary, the combination of Dijkstra's algorithm 
and ACO algorithm in reservoir irrigation scheduling 
design has achieved remarkable results, but there are still 
some areas that need to be improved and optimized. 
Future research can further explore how to optimize the 
model parameters and algorithms to improve the 
prediction accuracy. Meanwhile, it is also necessary to 
pay attention to the coordination of irrigation and power 
generation demand, as well as the cultivation and 
introduction of talents in the operation and management 
of the system. Through continuous exploration and 
practice, the system will play a greater role in the 
sustainable development of agriculture and energy. 
5 Conclusion 
To solve the problems of low resource utilization 
efficiency in traditional reservoir irrigation operation 
systems, in-depth research was conducted on the 
combination of Dijkstra algorithm and ACO in reservoir 
irrigation operation design. By effectively combining 
these two algorithms, the constructed irrigation operation 
system model could better handle multi-objective and 
multi-constraint optimization problems. The results 
showed that in the application of the reservoir irrigation 
operation system model, the average predicted water 
demand hydrograph by the system was 801.6L/s-1, which 
was 15.3L/s-1 different from the actual water demand 
hydrograph. At the same time, in the system model, the 
comprehensive operation method for irrigation and power 
generation had an irrigation water supply of 415 million 
m3, an annual irrigation guarantee rate of 85.32%, a 
water supply rate during the destruction period of 62.37%, 
a reduced water supply guarantee rate of 100%, an 
average annual power generation of 108.67 million kW·h, 
and a guaranteed power quantity of 71.529 million kW·h. 
This indicated that the reservoir irrigation operation 
system placed more emphasis on the synergistic benefits 
of irrigation and power generation. The system not only 
focused on the rational allocation of water resources, but 
also focused on achieving the dual goals of agricultural 
production and power supply through optimized 
operation. It could achieve rational utilization of water 
resources and improve irrigation efficiency, providing 
strong support for sustainable development of agriculture 
and energy. Meanwhile, the practical impacts of hybrid 
methods in actual reservoir management scenarios were 
categorized into three aspects: efficiency, resource 
allocation, and accuracy. The first was the computational 
efficiency. Hybrid methods could generate a large 
number of feasible solutions in a short period of time, 
thus greatly speeding up the search and optimization 
process. In the actual reservoir management, the 
scheduler could get the scheduling solution faster, which 
improved the efficiency and response time. The second 
was resource allocation. Hybrid method could give 
multi-objective constraint analysis, which helped to 
reduce the waste of water resources and improved the 
utilization efficiency in practical applications. The last 
was prediction accuracy. In practical application, higher 
prediction accuracy could help dispatchers more 
accurately grasp the trend of changes in irrigation 
demand and power generation demand, which provided 
strong support for the development of scientific 
scheduling programs. Although the research achieved 
significant results, there are still certain shortcomings. 
The exploration of uncertainty factors in the process of 
reservoir irrigation operation is not yet sufficient for input, 
and the next step is to conduct in-depth comprehensive 
evaluation of the uncertainty factors to improve the 
performance of the system model. 
References 
[1] V. Azamipour, N. Misaghian, and M. Assareh, 
"Multi-level optimization of reservoir scheduling using 
multi-resolution wavelet-based up-scaled models," 
Natural Resources Research, vol. 29, no. 2, pp. 
2103-2125, 2020. 
https://doi.org/10.1007/s11053-019-09538-w 
[2] J. Bang, J. Y. Choi, and P. Yoon, "Assessing irrigation 
water supply from agricultural reservoir using 

184   Informatica 48 (2024) 171 –184                                                                     B. Jia 
automatic water level data of irrigation canal," Journal 
of the Korean Society of Agricultural Engineers, vol. 
63, no. 1, pp. 27-35, 2021. 
https://doi.org/10.5389/KSAE.2021.63.1.027 
[3] S. Choudhuri, S. Adeniye, and A. Sen, "Distribution 
alignment using complement entropy objective and 
adaptive consensus-based label refinement for partial 
domain adaptation," Artificial Intelligence and 
Applications, vol. 1, no. 1, pp. 43-51, 2023. 
https://doi.org/10.47852/bonviewAIA2202524 
[4] S. Alshammrei, S. Boubaker, and L. Kolsi, "Improved 
Dijkstra algorithm for mobile robot path planning and 
obstacle avoidance," Computers, Materials and 
Continua, vol. 72, no. 7, pp. 5939-5954, 2022. 
https://doi.org/10.32604/cmc.2022.028165 
[5] A. Mahdavi, B. Shirazi, and J. Rezaeian, "Toward a 
scalable type-2 fuzzy model for resource-constrained 
project scheduling problem," Applied Soft Computing, 
vol. 100, no. 13, pp. 1-21, 2021. 
https://doi.org/10.1016/j.asoc.2020.106988 
[6] N. Chen, Z. Wang, R. He, J. Jiang, F. Cheng, and C. Han, 
"Efficient scheduling map** algorithm for row parallel 
coarse-grained reconfigurable architecture," Tsinghua 
Science and Technology, vol. 26, no. 5, pp. 724-735, 
2021. https://doi.org/10.26599/TST.2020.9010035 
[7] A. Priya and S. K. Sahana, "An optimized modelling and 
simulation on task scheduling for multi-processor 
system using hybridized ACO-CVOA," Journal of 
Information Science and Engineering, vol. 38, no. 5, pp. 
895-907, 2022. 
https://doi.org/10.6688/JISE.202209_38(5).0001 
[8] P. I. Bakanos and K. L. Katsifarakis, "Optimizing current 
and future hydroelectric energy production and water 
uses of the complex multi-reservoir system in the 
Aliakmon River, Greece," Energies, vol. 13, no. 24, pp. 
1-23, 2020. https://doi.org/10.3390/en13246499 
[9] A. Boluwade, "Remote sensed-based rainfall estimations 
over the East and West Africa regions for disaster risk 
management," ISPRS Journal of Photogrammetry and 
Remote Sensing, vol. 167, no. 6, pp. 305-320, 2020. 
https://doi.org/10.1016/j.isprsjprs.2020.07.015 
[10] H. Waluyadi, P. T. Juwono, R. Widandi Soetopo, L. M. 
Limantara, and D. Legono, "Optimization model of 
reservoir operating pattern based on the adaptive 
inflow for climate change," Journal of Southwest 
Jiaotong University, vol. 56, no. 2, pp. 491-511, 2021. 
https://doi.org/10.35741/issn.0258-2724.56.2.40 
[11] L. W. Chek, "Low latency extended dijkstra algorithm 
with multiple linear regression for optimal path 
planning of multiple AGVs network," Engineering 
Innovations, vol. 6, no. 9, pp. 31-36, 2023. 
https://doi.org/10.4028/p-t122xr 
[12] P. Li, C. Yang, R. Wang, and S. Wang, "A 
high-efficiency, information-based exploration path 
planning method for active simultaneous localization 
and mapping," International Journal of Advanced 
Robotic Systems, vol. 17, no. 1, pp. 1-14, 2020. 
https://doi.org/10.1177/1729881420903207 
[13] Y. Wang, J. Chen, W. Ning, H. Yu, and C. Chen, "A 
time-sensitive network scheduling algorithm based on 
improved ant colony optimization," Alexandria 
Engineering Journal, vol. 60, no. 1, pp. 107-114, 2021. 
https://doi.org/10.1016/j.aej.2020.06.013 
[14] W. Yuan, X. Yu, C. Su, D. Yan, and Z. Wu, "A 
multi-timescale integrated operation model for 
balancing power generation, ecology, and water supply 
of reservoir operation," Energies, vol. 14, no. 1, pp. 
1-21, 2020. https://doi.org/10.3390/en14010047 
[15] G. Yang, M. Giuliani, and S. Galelli, "Valuing the 
codesign of streamflow forecast and reservoir 
operation models," Journal of Water Resources 
Planning and Management, vol. 149, no. 8, pp. 1-11, 
2023. 
https://doi.org/10.1061/JWRMD5.WRENG-6023 
[16] H. Y. Kim, W. H. Nam, Y. S. Mun, N. K. Bang, and H. 
J. Kim, "Estimation of irrigation return flow on 
agricultural watershed in Madun reservoir," Journal of 
the Korean Society of Agricultural Engineers, vol. 63, 
no. 2, pp. 85-96, 2021. 
https://doi.org/10.5389/KSAE.2021.63.2.085 
[17] C. M. Rodrigues, M. Moreira, and R. C. Guimarães, 
"Reservoir evaporation in a Mediterranean climate: 
comparing direct methods in Alqueva Reservoir, 
Portugal," Hydrology and Earth System Sciences, vol. 
24, no. 12, pp. 5973-5984, 2020. 
https://doi.org/10.5194/hess-24-5973-2020 
[18] M. Rojas-Santiago, S. Muthuswamy, and M. Hulett, 
"An ACO algorithm for scheduling a flow shop with 
setup times," International Journal of Industrial and 
Systems Engineering, vol. 36, no. 1, pp. 98-109, 2020. 
https://doi.org/10.1504/IJISE.2020.109123 
[19] B. T. Agyeman, S. R. Sahoo, J. Liu, and S. L. Shah, 
"An LSTM-based mixed-integer model predictive 
control for irrigation scheduling," The Canadian 
Journal of Chemical Engineering, vol. 101, no. 6, pp. 
3362-3381, 2023. https://doi.org/10.1002/cjce.24764 
[20] S. K. Roy, S. Misra, N. S. Raghuwanshi, and S. K. Das, 
"AgriSens: IoT-based dynamic irrigation scheduling 
system for water management of irrigated crops," IEEE 
Internet of Things Journal, vol. 8, no. 6, pp. 5023-5030, 
2020. https://doi.org/10.1109/JIOT.2020.3036126 
[21] L. Lucky and A. S. Girsang, "Hybrid nearest neighbors 
ant colony optimization for clustering social media 
comments," Informatica, vol. 44, no. 1, pp. 63-74, 
2020. https://doi.org/10.31449/inf.v44i1.2672 
[22] H. Wang, "Research on data transmission optimization 
of communication network based on reliability 
analysis," Informatica, vol. 44, no. 3, pp. 361-366, 
2020. https://doi.org/10.31449/inf.v44i3.3280