https://doi.org/10.31449/inf.v47i2.3958 Informatica 47 (2023) 253–260 253 
Improved Genetic Algorithm in Multi-objective Cargo Logistics 
Loading and Distribution 
 
Zhilin He 
Chengdu Agriculture college, Economic trade department, Chengdu, Sichuan, 611130, China 
E-mail: zhilinhe3@163.com  
 
Keywords: improved genetic algorithm, material distribution, path planning, optimization objectives 
 
Received: February 2, 2022 
 
To solve the problem of material distribution path planning in a production workshop, this paper 
proposes research on multi-objective cargo logistics loading and distribution based on an improved 
genetic algorithm. This paper improves the genetic algorithm to solve the problem (P), that is, the 
evolution model based on the genetic algorithm draws lessons from the coding mode of the genetic 
algorithm, and uses the row insertion method to obtain the initial population. The improved genetic 
algorithm is better than the traditional genetic algorithm. The rapid development of railway 
transportation towards high speed, high density, and heavy load has led to even higher requirements for 
the safety of railway signal equipment. The safety of railway signal equipment is an important part of 
ensuring railway traffic safety, thus, it is very necessary to study a system that can diagnose the fault of 
railway signal equipment according to the actual situation. This article utilizes the genetic algorithm of 
artificial intelligence for investigating the loading and distribution of logistics in transportation. It is 
demonstrated that genetic algorithm integration is an effective method to improve the performance of 
logistic distribution model. The convergence speed of the improved genetic algorithm is fast, and it 
shows a stable upward trend with the increase of the number of iterations. 
 
Povzetek: Predlagan je izboljšan genetski algoritem za učinkovito logistično natovarjanje in 
distribucijo, kar izboljšuje uspešnost in hitrost logistične distribucije. 
 
1 Introduction  
With the rapid development of science and 
technology and the acceleration of the process of global 
economic integration, the logistics activities in the daily 
business activities of enterprises play a more and more 
important role in the global economic development, and 
their impact on all aspects of global economic activities 
is becoming more and more obvious [1]. Especially with 
the development of e-commerce websites, logistics has 
gradually become an important competitive field for 
enterprises [2]. Logistics distribution refers to integrated 
and systematic management that coordinates with the 
daily business activities of enterprises such as 
production, procurement, supply, and sales, and 
integrates the corresponding logistics activities such as 
distribution, handling, loading and unloading, storage, 
transportation, packaging, and information transmission. 
Its purpose is to provide customers with the best service 
at the lowest possible cost, to improve the overall 
economic benefits of the enterprise and enhance the 
overall competitive level of the enterprise [3].  
At present, in China's logistics industry, important 
decision-making problems such as distribution center 
location, transportation route, inventory control and 
cargo assembly scheme are still in the state of semi 
manual decision-making, and the technical support of the 
whole logistics activities lags behind the global average 
level [4]. Like the vast majority of different businesses, 
sharing is a major problem for logistics factories now, 
from Uber-style ways to deal with last-mile conveyance, 
to more operations specialist organizations and 
organizations at big business level, cooperation is re-
imagined in the entire SLN (sharing logistics network). 
By and large, various capabilities in SLN can be shared, 
from request assortment, pressing, arranging, to ship, 
stockpiling and conveyance, qualified conventional 
suppliers are allowed to go along with it and import their 
own accessible operations assets into asset pool for 
summon, and the coordinated suppliers perform brought 
together asset evening out to fulfil irregular strategies 
needs from different clients. Contrasted and the ordinary 
frameworks independently set-up, SLN with enormous 
clients prompts further developing individual asset 
productivity, defending by and large asset work, reducing 
transport expense, and saving conveyance time.  
For instance, asset evening out in SLN can further 
develop responsibility by changing the rundown of 
operations task start times and re-establishing the priority 
connection among undertakings intra-stage, and 
furthermore can keep up with the power in asset 
productivity considering benefit assignment and 
administration criticism between stage. One might say 
that multi-stage asset evening out among customary 
suppliers is particularly essential to asset the board in 
SLN, which is the obvious elements of Industry 3.5. 
Multi-stage asset evening out issue in sharing planned 
operations network is a multi-objective improvement 
issue, which is firmly non-deterministic polynomial hard 
in open circle climate.  

254   Informatica 47 (2023) 253–260                                                                                                                                             Z. He 
 
Figure 1: Material distribution process 
 
Especially in the e-commerce environment with 
huge transaction volumes and extremely fast transaction 
speed, the contradiction between information flow and 
logistics has seriously affected the service quality of 
enterprises. Therefore, through the research on the 
optimization of distribution path, we can clarify the work 
details of each link of logistics distribution in enterprise 
business activities, and further improve the relevant 
technical system, to improve the overall intelligent level 
of the logistics industry. It is assumed that each 
operator's initial position is in the material supermarket. 
When the operator completes all batching tasks, he 
returns to the material supermarket. Dispatch the 
minimum number of operators to load the material truck 
according to the assigned task and send it to each 
workstation in turn until all materials are delivered and 
returned to the material supermarket. The material 
distribution process is shown in Figure 1 [5]. 
The rest of this article is organized as: Section 2 
presents the related works in various domains. Section 3 
consists of multi-objective mathematical model and its 
solution using genetic algorithm. Results and analysis are 
discussed in section 4 followed by concluding remarks in 
section 5. 
 
2 Related work 
Aiming at this research problem, Yun and Li 
changed the constraints in the distribution path and 
analyzed the impact of each index on the path [6]. Feng 
et al. proposed an optimization model aiming at the 
shortest distribution path and the least distribution times 
[7]. Sheng et al. used a hybrid particle swarm 
optimization algorithm to solve the material distribution 
model aiming at material transportation cost, 
transportation time, and line inventory [8]. Abbasi et al. 
proposed a material distribution method focusing on 
stations, and established a material flow path 
optimization model with the optimization objective of 
minimizing the total distribution time [9]. Wang et al. 
mainly establish models for VRP problems in two 
processes of auxiliary materials and replenishment and 
use a genetic algorithm to solve them with the goal of 
path optimization [10]. Saric et al. studied the just-in-
time distribution of materials under the condition of 
limited distribution vehicles and solved it with an 
improved genetic algorithm and three-stage heuristic 
algorithm [11]. Mazhari et al. expounded on the 
problems existing in the material distribution process of 
the automobile assembly line and put forward the 
material distribution path optimization model with a time 
window with the goal of the shortest distribution distance 
[12]. Zan et al. proposed an optimization model to 
minimize distribution cost and solved the model by using 
dynamic programming and simulated annealing genetic 
algorithm [13]. Minaei et al. proposed an M- VRPTW 
research model based on the VRPTW problem [14]. 
Grisales et al. proposed a VRPTW mathematical model 
with time window constraints based on multi-supply 
points and place dependence and further summarized the 
research progress of VRPTW [15]. Based on the current 
research, this paper proposes the research of multi-
objective cargo logistics loading and distribution based 
on an improved genetic algorithm.  
A technique which considers wavelet frames for 
micropolar fluid flow is used for high mass transfer [16]. 
The vibration over the sandwich plates of laminated 
skew is studied through finite element [17]. The 
numerical simulation based on space time fractional 
equation are evaluated [18]. The proposed work can 
further be extended by using integration approaches of 
Artificial Intelligence and Machine learning as studied 

Improved Genetic Algorithm in Multi-objective Cargo Logistics…                                          Informatica 47 (2023) 253–260     255 
from several studies [19-21]. This paper improves the 
genetic algorithm to solve the problem (P), that is, the 
evolution model based on the genetic algorithm draws 
lessons from the coding mode of the genetic algorithm, 
and uses the row insertion method to obtain the initial 
population. In the crossover operation, the narrow gene 
similarity is used to distinguish the chromosome 
similarity, and the double variation rate is added to the 
mutation operation in the evolution process. MATLAB is 
used to realize the algorithm, and compared with the 
traditional genetic algorithm to verify the feasibility and 
effectiveness of the proposed model and algorithm. 
 
3 Mathematical model establishment 
This section includes the description of model 
establishment including variables and providing solution 
based on genetic algorithm.  
3.1 Decision variables 
 



=
Otherwise; , 0
J; station   to I station  from  tool Moving , 1
ijk
x
 
(1) 



=
Otherwise; , 0
K;  tool handling by  serviced is I Station , 1
ik
y
 
(2) 
3.2 Multi-objective mathematical model 
According to the problems, the following 
mathematical models are constructed: 
 
A. Optimization objectives 
Minimize total distance traveled: 
 
ijk
K
k
N
i
N
j
ij
x d d
  
= = =
=
1 0 0
min
 
(3) 
Minimize vehicle usage: 
 
= =
=
N
i
K
k
jk
x v
0 1
0
min 
(4) 
Average satisfaction of the largest chemical 
industry: 
( )
i
N
i
i
t u
N
u

=
=
1
1
max 
(5) 
Maximize vehicle capacity: 
KQ
q
l
N
i
i 
=
=
1
max 
(6) 
B. Constraints 
  N i LT t ET
i i i
,..., 2 , 1 ,    (7) 
  K k Q y q
ik
N
i
iQ
,..., 2 , 1 ,
1
 

=
 (8) 
  N i y
K
k
ik
,..., 2 , 1 , 1
1
 =

=
 (9) 
  K k y x
jk
N
I
ijk
,..., 2 , 1 ,
0
 =

=
 (10) 
  K k y x
jk
N
j
ijk
,..., 2 , 1 ,
0
 =

=
 
(11) 
  K k x
N
i
k i
,..., 2 , 1 , 1
0
0
 =

=
 (12) 
  K k x
N
j
ik
,..., 2 , 1 , 1
0
0
 =

=
 
(13) 
  n R R x
R
i
R
j
ijk
,..., 2 , 1 , 1
1 1
 − 
 
= =
 
(14) 
Where 𝑛 is the number of jobs; K is the number of 
distribution tools; 𝑄 is the carrying capacity; 𝑄 𝑖 is the 
demand of station 𝑖 ; 𝐷 𝑖𝑗
 is the distance from stations 𝑖 
and 𝑗 ; 𝐷 is the service time at station 𝑖 ; 𝑇 𝑖𝑗
 is the travel 
time from station 𝑖 to station 𝑗 ; 𝑊 𝑖 is the waiting time at 
station 𝑖 ; 𝑡 𝑖 is the time to start service station 𝑖 . Equation 
(3) ~ equation (6) represents the objective function, 
which is the driving distance of distribution tools, 
number of used tools, station satisfaction, and carrying 
rate in turn; Equation (7) means that the service shall not 
be started outside the time window; Equation (8) 
represents the carrying capacity constraint; Equation (9) 
indicates that each station has and only has one 
distribution tool service; Equations (10) to (11) represent 
the relationship between two variables 𝑥 𝑖𝑗𝑘 and 𝑦 𝑖𝑘
; 
Equations (12) to (13) indicate that the distribution tools 
leave the warehouse and finally drive back to the 
warehouse; Equation (14) is a branch elimination 
constraint [22]. 
3.3 Solving the model with an improved 
genetic algorithm 
The material distribution path planning problem in 
the production workshop is an NP problem. At present, 
many heuristic algorithms have their defects in solving 
multi-objective and multi-constraint optimization 
problems, such as slow convergence, premature 
convergence, unbalanced global optimization ability, and 
local search ability. Therefore, this paper improves the 
genetic algorithm to solve the problem (P), that is, based 
on the evolution model of the genetic algorithm, draws 

256   Informatica 47 (2023) 253–260                                                                                                                                             Z. He 
lessons from the coding mode of the genetic algorithm, 
and uses the row insertion method to obtain the initial 
population. In the crossover operation, the narrow gene 
similarity is used to distinguish the similarity of 
chromosomes, and the double mutation rate is added to 
the mutation operation in the evolution process [23]. 
3.4 Construct chromosomes to produce the 
initial population 
The quality of initial population generation 
determines the starting point of algorithm search. The 
high-quality population can accelerate the convergence 
of the algorithm and improve the solution efficiency [24]. 
When there is no feasible insertion position in the current 
distribution tool path, a new distribution tool is opened, 
and this paper improves it on this basis, as described in 
step 4. 
Definition 1: The optimal insertion position is in the 
set of feasible insertion positions of the current path, and 
the initial feasible population generation steps are as 
follows: 
Step 1: Randomly arrange 𝑁 stations to obtain the 
station sequence set 𝑋 = {𝑥 1
, 𝑥 2
, … , 𝑥 𝑁 }, then assign it to 
the vehicle from left to right, and finally initialize the 
distribution path number Car_Code=0 ； 
Step 2: Set car_Code = Car_Code + 1, the newly 
opened delivery tool, and its path number is Car_Code, 
arrange the leftmost station in the current x to the new 
path, and delete the station from 𝑋 ; 
Step 3: Judge whether 𝑋 is empty. If yes, go to step 
5; Otherwise, go to step 4. 
Step 4: Take the leftmost station in the current set X 
and judge whether there is a feasible insertion position in 
the current distribution tool path. If it exists, insert it into 
the best insertion position of the current distribution tool 
path, and turn to step 3; If it does not exist, judge 
whether it exists from small to large according to the 
path number of the distribution tool. If it exists, insert it 
into the best insertion position of the path, and turn to 
step 3, otherwise turn to step 2. 
Step 5: Calculate the best start service time of the 
generated feasible chromosome to maximize the average 
station satisfaction of the chromosome. 
Step 6: Repeat steps 2 to 5, and stop when the number of 
feasible chromosomes reaches the specified number. 
Among them, the feasible insertion location and the best 
service start time are determined as follows: 
 
i. Determination of feasible insertion position 
If the feasible distribution path of an established 
distribution tool is 𝐵 and the start service time at the 
station 𝑥 𝑖 is 𝑡 𝑥𝑖
, the maximum delay time for the 
distribution tool to arrive at the station is: 
 
( )
( )





−  





 + −
= + −
=
+ −
−
1 1 ,
max
,
min
,
max
1
p i
x pone
W t LT
p i W t LT
x pone
i
xi xi xi
xi xi xi
i
 
(15) 
If a station 𝑥 𝑠 is inserted after the 𝑖 𝑡 ℎ
 station in the 
feasible path, the path is still feasible, then condition 1 
capacity constraints: 
Q q q
s
p
i
xi
 +

=1
 
(16) 
Condition 2-time constraints, the time when the 
delivery tool reaches the station s shall not be later than 
the lower limit of the time window of the station, that is, 
when 𝑡 𝑥𝑖
+ 𝑇 𝑥𝑖
+ 𝑇 𝑥𝑖 ,𝑠 ≤ 𝐿𝑇
𝑠 and 𝑖 = 0,  𝑡 𝑥𝑖
= 𝑇 𝑥𝑖
= 0, 
𝑇 𝑥 1,𝑠 = 𝑇 0,𝑠 . When 𝑖 ≤ 𝑃 − 1 the delivery tool must meet 
the following requirements when arriving at the station 
𝑋 𝑖 +1
: 
 
max{𝑡 𝑥𝑖
+ 𝑇 𝑥𝑖
+ 𝑇 𝑥𝑖 ,𝑠 , 𝐸𝑇
𝑠 } + 𝑇 𝑠 + 𝑇 𝑠 ,𝑥𝑖 +1
≤ 
(17) 
𝑡 𝑥𝑖 +1
+ max _𝑝𝑜𝑛𝑒 (𝑥 𝑖 +1
) − 𝑊 𝑥𝑖 +1
 
ii. Determination of the best service start time 
In the feasible path obtained by the above operation, 
the comprehensive satisfaction of the station is not the 
best. Based on the research on the determination method 
of the best start service time, this paper adjusts the time 
from the last station of the distribution tool path, omits 
the determination of the non-push able part, and 
improves the calculation of the maximum deferrable 
amount. 
The calculation steps of the best service start time 
are as follows: 
Step 1: Assume that the distribution path of a 
distribution tool is (𝑥 1
, 𝑥 2
, … , 𝑥 𝑝 ). 
Step 2: Divide the delivery tool path into several 
segments (𝑥 𝑖 , 𝑥 𝑖 +1
, 𝑥 𝑖 +2
, … , 𝑥 𝑗 ), meet condition 𝑊 𝑥𝑖 +1
=
𝑊 𝑥𝑖 +2
= ⋯ = 𝑊 𝑥𝑗
= 0, when 𝑖 ≠ 1 ，𝑊 𝑥𝑖
≠ 0; 𝑊 𝑥𝑗
<
𝑝 , 𝑊 𝑥𝑗 +1
≠ 0. 
Step 3: Take the last segment of the distribution 
path and set it as (𝑥 1
, 𝑥 2
, … , 𝑥 𝑚 ), Adjust the service start 
time of the station 𝑥 𝑚 , 𝑥 𝑚 −1
, … , 𝑥 2
, 𝑥 1
 in sequence; 
Under the condition of meeting the time constraint, the 
maximum delay to station 𝑥 𝑠 (𝑠 = 1, 2, … , 𝑚 ) is 𝑡 0
=
max{𝑡 ′
|∑ 𝑢 𝑥𝑛
′
(𝑡 𝑥𝑛
+ 𝑡 ′
)
𝑚 𝑛 =𝑠 ≥ 0}. 
Step 4: Take the previous segment of the adjusted 
segment in the current distribution tool path and adjust it 
according to the method in step 3; If the segment forms a 
new segment with the latter segment in the current 
distribution tool path, readjust the newly formed 
segment; Otherwise, go to step 5. 
Step 5: Continue to repeat step 4 until the service 
start time of all stations in the distribution tool path is 
determined. 

Improved Genetic Algorithm in Multi-objective Cargo Logistics…                                          Informatica 47 (2023) 253–260     257 
3.5 Fitness function 
The fitness of each individual is calculated 
according to the optimization target value and weight 
parameter. The solving function is as follows: 
 
( )
4 , 3 , 2 , 1 , 0 , 1
1 1
4
1
max
4
max
3
max
2
max
1
=  =








− +








− + + =

=
i a
l
l
u
u
v
v
d
d
h f
i
i
i
k k k k
k

   
 
(18) 
Where, ℎ
𝑘 represents chromosome 𝐾 ; 𝑑 𝑘 , 𝑣 𝑘 , 𝑢 𝑘 , 𝑙 𝑘 
represent the driving distance of chromosome ℎ, the total 
number of distribution tools used, the average 
satisfaction of stations, and the carrying rate of 
distribution tools; 𝑑 𝑚𝑎𝑥
, 𝑣 𝑚𝑎𝑥
, 𝑢 𝑚𝑎𝑥
, 𝑙 𝑚𝑎𝑥
 represents the 
maximum travel distance, the maximum number of 
distribution tools used, the maximum average satisfaction 
of stations, and the maximum carrying rate of 
chromosomes in the current population; 𝛼 𝑖 (𝑖 = 1, 2, 3, 4) 
is the weight coefficient [25]. 
 
4 Results and analysis 
This section illustrates the result and analysis of 
proposed system using genetic algorithm. The distance 
between stations is measured along with the driving 
schedule which is presented in this section. 
In material distribution in the production workshop, 
there are 9 stations and a warehouse, and the carrying 
capacity of the distribution tool is 12 units. The service 
time, reservation time, and demand of each station are 
shown in Table 1, and the driving time and distance 
between each station are shown in Tables 2 and 3; 
Improve the basic parameters of the genetic algorithm 
and take the population size pop_Size = 100, number of 
iterations Max_Gen = 200, selection probability 𝑃 𝑥 =
0.8, crossover probability 𝑃 𝑐 = 0.8, double mutation 
probability Local_Pm = 0.1 and Global_Pm=0.2.  
 
Table 1: Information about each station 
Station 
number 
1 2 3 4 5 6 7 8 9 
Service 
time/min 
3.1 5.1 7.8 5.2 4.2 6.9 7.6 6.2 5.5 
Reservati
on time / 
min 
33,3
7,39 
5.7.4.1
0 
26,27,
30 
18,20,
22 
15,18,
21 
31,34,
36 
9,11,1
2 
22,25,
27 
29,33,
35 
Demand/
unit 
3.8 5.2 2.7 2.8 4.4 4.8 5.7 2.7 4.7 
 
Table 2: Distance between stations and Driving Schedule 
(stations 1-4) 
Station 
number 
Driving time/min 
Warehouse 
0 
Station 
1 
Station 
2 
Station 
3 
Station 
4 
Driving 
distance/m 
0 0 10 5 9 11 
1 23.43 0 6 5 7 
2 27.12 24.10 0 9 7 
3 12.29 16.54 15 0 6 
4 26.49 36.59 18.49 21.07 0 
5 11.59 33.31 30.21 20 21.56 
6 7.11 29.30 32.42 18. 65 28.69 
7 7.79 26.79 34.39 19.51 33.49 
8 21.89 25.31 43.59 28.89 47.79 
9 20.11 7.89 29.79 18.35 39.80 
 
According to the above algorithm, Matlab 
simulation is used to calculate under different weight 
settings. The results are shown in Table 4. 
It can be seen from Table 4 that the setting of 
optimization target weight parameters has a great impact 
on the experimental results [26]. d. L is closely related to 
v, that is, d increases with the increase of v, and l 
decreases with the increase of 0; u is opposite to d, v and 
l, that is, the increase of u is at the cost of the increase of 
d and v and the decrease of L [21]. 
  
Table 3: Distance between stations and Driving Schedule 
(stations 5-9) 
Station 
number 
Driving time/min 
Station 5 Station 6 Station 7 Station 8 Station 9 
Driving 
distance/m 
0 5 11 7 10 10 
1 8 6 8 11 10 
2 9 12 7 11 11 
3 6 11 6 6 9 
4 8 6 11 9 10 
5 0 9 10 12 9 
6 9.59 0 10 10 9 
7 14.69 6.15 0 5 6 
8 31.39 22.39 17.19 10 10 
9 31.69 25.79 21.89 17.51 0 
 
Table 4: Simulation results 
Weight setting   Experimental result 
𝜶 𝟏 𝜶 𝟐 𝜶 𝟑 𝜶 𝟒 d v u l 
0.7 0.1 0.1 0.1 209.31 5 0.849 0.621 
0.1 0.7 0.1 0.1 227.53 4 0.679 0.767 
0.1 0.1 0.7 0.1 243.31 6 0.951 0.521 
0.1 0.1 0.1 0.7 231.21 4 0.761 0.776 
 
It can be seen from Table 4 that the setting of 
optimization target weight parameters has a great impact 
on the experimental results [20]. d. L is closely related to 
v, that is, d increases with the increase of v, and l 
decreases with the increase of 0; u is opposite to d, v and 
l, that is, the increase of u is at the cost of the increase of 
d and v and the decrease of L [27]. 
To prove the effectiveness of the improved genetic 
algorithm proposed in this paper, the improved genetic 
algorithm takes the weights 𝛼 1
= 1, 𝛼 2
= 1 and 𝛼 3
= 1 
respectively, which is compared with the corresponding 
single objective solution of the traditional genetic 
algorithm (𝛼 4
= 1 is not analyzed here because l is 
inversely proportional to v). The traditional genetic 
algorithm is used to calculate this example. The initial 
population size is 100, the selection probability and 

258   Informatica 47 (2023) 253–260                                                                                                                                             Z. He 
crossover probability are 0.8, and the mutation 
probability is 0.1 [28]. 
As shown in Figure 2, when 𝛼 1
= 1, the improved 
genetic algorithm shows a stable downward trend after 
30 generations and converges to 55 generations [29]. 
However, the convergence speed of the traditional 
genetic algorithm is very slow in the middle and late 
stages, and it does not begin to converge until generation 
126. As shown in Figure 3, when 𝛼 2
= 1, the improved 
genetic algorithm has no fluctuation. From the whole 
image, the two have a downward trend, that is, the 
connecting line between the starting point and the 
convergence point. The slope of the improved genetic 
algorithm is significantly greater than that of the 
traditional genetic algorithm [30].  
 
 
Figure 2: Comparison of travel distance convergence of 
distribution tools. 
 
 
Figure 3: Convergence comparison of usage number of 
distribution tools. 
 
As shown in Figure 4, when 𝛼 3
= 1, the 
convergence speed of the improved genetic algorithm is 
also fast, showing a stable upward trend with the increase 
in the number of iterations [31]. To sum up, from the 
convergence diagrams in Figures 2, 3, and 4, it is obvious 
that the improved genetic algorithm is superior to the 
traditional genetic algorithm [32]. 
 
 
Figure 4: Convergence comparison of station average 
satisfaction. 
In the crossover operation, the narrow gene 
similarity is used to distinguish the chromosome 
similarity, and the double variation rate is added to the 
mutation operation in the evolution process. The basic 
parameters of the genetic algorithm are improved and the 
population size pop is taken_ size = 100, number of 
iterations Max_ gen = 200, the selection probability 𝑃 𝑥 =
0.8, crossover probability 𝑃 𝑐 = 0.8, double mutation 
probability Local_ Pm = 0.1 and Global_ Pm=0.2. 
Matlab simulation is used to calculate under different 
weight settings. When 𝛼 1
= 1, the improved genetic 
algorithm shows a stable downward trend after 30 
generations and converges to 55 generations; However, 
the convergence speed of the traditional genetic 
algorithm is very slow in the middle and late stages, and 
it does not begin to converge until generation 126. When 
𝛼 2
= 1, the improved genetic algorithm has no 
fluctuation. From the whole image, we can see the 
downward trend between the two, that is, the connection 
between the starting point and the convergence point. 
The slope of the improved genetic algorithm is 
significantly greater than that of the traditional genetic 
algorithm. When 𝛼 3
= 1, the convergence speed of the 
improved genetic algorithm is fast, and it shows a stable 
upward trend with the increase in the number of 
iterations. 
 
5 Conclusions 
In this paper, based on the research of multi-
objective cargo logistics loading and distribution based 
on an improved genetic algorithm, a multi-objective 
mathematical programming model under the condition of 
fuzzy station reservation time is constructed, and an 
improved genetic algorithm is proposed to solve the 
model. In the algorithm design, the feasible insertion 
method is used to generate the initial population to speed 

Improved Genetic Algorithm in Multi-objective Cargo Logistics…                                          Informatica 47 (2023) 253–260     259 
up the convergence speed. The concept of narrow gene 
similarity is proposed to avoid the cross operation of 
individuals with high similarity. At the same time, the 
double mutation mechanism is added to the mutation 
process to control the search range and convergence 
speed of the global solution space. Finally, the algorithm 
is implemented by MATLAB and compared with the 
traditional genetic algorithm to verify the feasibility and 
effectiveness of the proposed model and algorithm. The 
next research will consider the dynamic changes in 
workstation material requirements. In future work this 
research, delicate time window should to be thought of as 
because of the impediment of difficult time window in 
this paper, which can recognize the distinction among 
calculations. Moreover, the double effect of vulnerability 
from resource supply and client request is one more 
worry in proposed system. 
 
References  
[1] Wang, X., Chang, J., Meng, X., & Wang, Y . (2017). 
Research on multi-objective operation based on 
improved NSGA-II for lower Yellow River. J. 
Hydraul. Eng, 48, 135-145.  
10.13243/j.cnki.slxb.20151361 
[2] Geng, K., Ye, C., & Liu, L. (2020). Research on 
multi-objective hybrid flow shop scheduling 
problem with dual resource constraints using 
improved memetic algorithm. IEEE Access, 8, 
104527-104542.  
10.1109/ACCESS.2020.2999680  
[3] Mei, K., Li, D., Yoon, S. W., & Ryu, J. H. (2016). 
Multi-objective optimization of collation delay and 
makespan in mail-order pharmacy automated 
distribution system. The International Journal of 
Advanced Manufacturing Technology, 83(1), 475-
488.  
https://doi.org/10.1007/s00170-015-7555-7 
[4] Arabzad, S. M., Ghorbani, M., & Tavakkoli-
Moghaddam, R. (2015). An evolutionary algorithm 
for a new multi-objective location-inventory model 
in a distribution network with transportation modes 
and third-party logistics providers. International 
Journal of Production Research, 53(4), 1038-1050.  
https://doi.org/10.1080/00207543.2014.938836 
[5] Zhang, L. X. (2019). Research on similarity 
measurement of video motion images based on 
improved genetic algorithm in paper 
industry. Paper Asia, 2(1), 135-138.  
10.1155/2022/3936958 
[6] Yun, Y ., & Li, Y . (2016). An Improved Nonlinear 
Multi-Objective Optimization Problem Based on 
Genetic Algorithm. International Journal of Hybrid 
Information Technology, 9(7), 361-372. (2016), pp. 
361-372  
http://dx.doi.org/10.14257/ijhit.2016.9.7.33 
[7] Feng, Q., Yang, F., Xu, X., Zhang, B., Ding, Y ., & 
Liu, Q. (2019). Multi-objective optimization genetic 
algorithm for multi-point light focusing in 
wavefront shaping. Optics Express, 27(25), 36459-
36473.  
https://doi.org/10.1364/OE.27.036459 
[8] Sheng, W., Liu, K. Y ., Liu, Y., Meng, X., & Li, Y . 
(2014). Optimal placement and sizing of distributed 
generation via an improved nondominated sorting 
genetic algorithm II. IEEE Transactions on power 
Delivery, 30(2), 569-578.  
10.1109/TPWRD.2014.2325938 
[9] Abbasi, M., Mohammadi Pasand, E., & Khosravi, 
M. R. (2020). Workload allocation in iot-fog-cloud 
architecture using a multi-objective genetic 
algorithm. Journal of Grid Computing, 18(1), 43-
56.  
https://doi.org/10.1007/s10723-020-09507-1 
[10] Wang, C., Feng, Y ., Ye, J., Luo, B., & Liu, K. 
(2017). Multi-objective parameters optimization of 
centrifugal slurry pump based on RBF neural 
network and NSGA-genetic algorithm. Transactions 
of the Chinese Society of Agricultural 
Engineering, 33(10), 109-115.  
[11] Šarić, M., Hivziefendić, J., & Konjić, T. (2019). 
Multi-objective DG allocation in a radial power 
distribution network for power loss reduction, 
voltage profile improvement and investment 
deferral. Elektrotehniski Vestnik, 86(5), 253-258.  
[12] Mazhari, S. M., Monsef, H., & Romero, R. (2015). 
A multi-objective distribution system expansion 
planning incorporating customer choice on 
reliability. IEEE Transactions on Power 
Systems, 31(2), 1330-1340.  
10.1109/TPWRS.2015.2430278 
[13] Zan, T. T. T., Gupta, P., Wang, M., Dauwels, J., & 
Ukil, A. (2018). Multi-objective optimal sensor 
placement for low-pressure gas distribution 
networks. IEEE Sensors Journal, 18(16), 6660-
6668.  
10.1109/JSEN.2018.2850847 
[14] Minaei, A., Sabzkouhi, A. M., Haghighi, A., & 
Creaco, E. (2020). Developments in multi-objective 
dynamic optimization algorithm for design of water 
distribution mains. Water Resources 
Management, 34(9), 2699-2716.  
https://doi.org/10.1007/s11269-020-02559-8 
[15] Grisales, L. F., Grajales, A., Montoya, O. D., 
Hincapie, R. A., Granada, M., & Castro, C. A. 
(2017). Optimal location, sizing and operation of 
energy storage in distribution systems using multi-
objective approach. IEEE Latin America 
Transactions, 15(6), 1084-1090.  
10.1109/TLA.2017.7932696  
[16] Kumbinarasaiah, S., & Raghunatha, K. R. (2021). A 
novel approach on micropolar fluid flow in a porous 
channel with high mass transfer via wavelet 
frames. Nonlinear Engineering, 10(1), 39-45. 
https://doi.org/10.1515/nleng-2021-0004 
[17] Dhotre, P. K., & Srinivasa, C. V . (2021). On free 
vibration of laminated skew sandwich plates: A 
finite element analysis. Nonlinear 
Engineering, 10(1), 66-76.  
https://doi.org/10.1515/nleng-2021-0006 
[18] Pedram, L., & Rostamy, D. (2021). Numerical 
simulations of stochastic conformable space–time 

260   Informatica 47 (2023) 253–260                                                                                                                                             Z. He 
fractional Kortewegde Vries and Benjamin–Bona–
Mahony equations. Nonlinear Engineering, 10(1), 
77-90.  
https://doi.org/10.1515/nleng-2021-0007 
[19] Wang, H., Hao, L., Sharma, A., & Kukkar, A. 
(2022). Automatic control of computer application 
data processing system based on artificial 
intelligence. Journal of Intelligent Systems, 31(1), 
177-192.  
https://doi.org/10.1515/jisys-2022-0007 
[20] Sun, L., Gupta, R. K., & Sharma, A. (2022). 
Review and potential for artificial intelligence in 
healthcare. International Journal of System 
Assurance Engineering and Management, 13(1), 
54-62.  
https://doi.org/10.1007/s13198-021-01221-9  
[21] Cai, Y ., & Sharma, A. (2021). Swarm intelligence 
optimization: an exploration and application of 
machine learning technology. Journal of Intelligent 
Systems, 30(1), 460-469.  
https://doi.org/10.1515/jisys-2020-0084 
[22] Kaur, R., & Kinger, S. (2014). Enhanced genetic 
algorithm-based task scheduling in cloud 
computing. International Journal of Computer 
Applications, 101(14), 1-6.  
10.5120/17752-8653 
[23] Abdelaziz, M. (2017). Distribution network 
reconfiguration using a genetic algorithm with 
varying population size. Electric Power Systems 
Research, 142, 9-11.  
https://doi.org/10.1016/j.epsr.2016.08.026 
[24] Wang, H. K., Chien, C. F., & Gen, M. (2015). An 
algorithm of multi-subpopulation parameters with 
hybrid estimation of distribution for semiconductor 
scheduling with constrained waiting time. IEEE 
Transactions on Semiconductor 
Manufacturing, 28(3), 353-366.  
10.1109/TSM.2015.2439054 
[25] Sun, Y ., Yen, G. G., & Yi, Z. (2018). Improved 
regularity model-based EDA for many-objective 
optimization. IEEE Transactions on Evolutionary 
Computation, 22(5), 662-678.  
10.1109/TEVC.2018.2794319 
[26] Ebrahimi, S., Ghorbani, B., & Vijayaraghavan, K. 
(2017). Optimization of catalyst distribution along 
PEMFC channel through a numerical two-phase 
model and genetic algorithm. Renewable 
Energy, 113, 846-854.  
https://doi.org/10.1016/j.renene.2017.06.067 
[27] Sankary, N., & Ostfeld, A. (2017). Scaled 
multiobjective optimization of an intensive early 
warning system for water distribution system 
security. Journal of Hydraulic Engineering, 143(9), 
04017025.  
https://doi.org/10.1061/(ASCE)HY .1943-
7900.0001317 
[28] Manjunath Patel, G. C., Krishna, P., & 
Parappagoudar, M. B. (2016). Modelling and multi-
objective optimisation of squeeze casting process 
using regression analysis and genetic 
algorithm. Australian Journal of Mechanical 
Engineering, 14(3), 182-198.  
https://doi.org/10.1080/14484846.2015.1093231 
[29] Santos, M. C., Machado, A. R., Barrozo, M. A. S., 
Jackson, M. J., & Ezugwu, E. O. (2015). Multi-
objective optimization of cutting conditions when 
turning aluminum alloys (1350-O and 7075-T6 
grades) using genetic algorithm. The International 
Journal of Advanced Manufacturing 
Technology, 76(5), 1123-1138.  
https://doi.org/10.1007/s00170-014-6314-5 
[30] Liang, C., Li, M., Lu, B., Gu, T., Jo, J., & Ding, Y . 
(2017). Dynamic configuration of QC allocating 
problem based on multi-objective genetic 
algorithm. Journal of Intelligent 
Manufacturing, 28(3), 847-855.  
https://doi.org/10.1007/s10845-015-1035-7  
[31] Nazarahari, M., Khanmirza, E., & Doostie, S. 
(2019). Multi-objective multi-robot path planning in 
continuous environment using an enhanced genetic 
algorithm. Expert Systems with Applications, 115, 
106-120.  
https://doi.org/10.1016/j.eswa.2018.08.008 
[32] Guo, Q. (2018). Research on the application of 
improved shuffled frog leaping algorithm in 
mechanical fault diagnosis. Academic Journal of 
Manufacturing Engineering, 16(2).