Wireless Pers Commun

https://doi.org/10.1007/s11277-017-5199-7

Design of Intelligent Warehouse Management System

Jia Mao1 • Huihui Xing1 • Xiuzhi Zhang2

Ó Springer Science+Business Media, LLC, part of Springer Nature 2018

Abstract With the continuous development and application of information technology in

the logistics industry, mobile applications, bar code, wireless communication, system
integration and other technologies are widely used in warehouse management. Focusing on
warehousing process management has become a new trend of informatization and lean
management of material warehouse management. In this paper, the management of dis-
tributed storage cloud model is discussed, and a variety of automation, intelligence and
information technology are utilized to effectively integrate the resources effectively. The
author studies the resource scheduling optimization problem in the system, proposes an
effective scheduling method, and initially realizes the intelligent warehouse management
system based on cloud model, which provides users with a good package management
service.

Keywords Intelligent warehouse  Management system  Model

1 Introduction

The development of modern logistics industry needs the support of information technology
[1]. Information technology makes the information system run through different enter-
prises, which greatly improves the efficiency of logistics. At the same time, it also provides
the conditions for the integration of enterprise demand, distribution and inventory man-
agement. Logistics informatization can be applied to every link of the logistics process,
which can be used for systematic control of a single link, such as transportation, storage
and procurement, and can also be applied to the whole logistics process. Overall, logistics

& Jia Mao

[email protected]
1
College of Transportation, Jilin University, Changchun, China
2
Institute of Mechanical Science and Engineering, Jilin University, Changchun, China

123
 J. Mao et al.

information is the use of modern information technology in the entire process of logistics
information collection, identification, classification, aggregation, transmission, sharing,
tracking and query. It can realize the effective communication and seamless connection of
the supply, demand, storage and transportation of materials, and construct a logistics
supply chain with both high speed and high efficiency and low cost advantage [2].
Therefore, the development of information technology has become the inevitable require-
ment of the development of the logistics industry, and the progress of information tech-
nology helps to improve the efficiency and quality of logistics delivery.

2 State of the Art

At present, in some large and medium-sized logistics companies with relatively high level
of informatization, E-commerce platform based on enterprise website are rare. Most of the
function of enterprise website are to promote the corporate image [3]. In addition, some of
the logistics warehousing information system has been completed. Its functions generally
include order management, warehousing management, financial management and trans-
portation management, but there are few achievements in the construction of warehouse
logistics network information [4]. In this context, many scholars have studied warehouse
management system. Studies have pointed out that the problems in the development of the
warehouse management system is mainly embodied in the following aspects: First, the
scale of warehouse resources in our country is large, but most of the warehouse man-
agement functions are not perfect, and the degree of information is low, resulting in low
logistics efficiency and increased inventory costs year by year. Second, enterprises do not
pay enough attention to warehouse management. They believe that warehouse information
cannot bring direct economic sources for enterprises, and it is not very helpful to improve
the efficiency of enterprises, so they are reluctant to invest in warehousing information
management [5]. Third, even if some enterprises build warehouse management system, due
to the ambiguity of the previous design ideas and the failure of the management system, the
warehouse management system has not reduced the workload of the warehouse manage-
ment personnel, but it has caused the tedious and complicated steps of the document input
[6]. The warehouse manager produced resistance to the system, leading to serious aban-
doned system. Fourth, it lacks support from policies and regulations. At present, logistics
warehousing management, construction standards, and the market order is not perfect.
There is no complete warehouse management laws and regulations for the logistics
industry to restrict the industry order [7].

3 Methodology

3.1 System Architecture

The structure of the intelligent warehouse management system is shown in Fig. 1, which is
mainly composed of mobile terminal system, application processing center, centralized
printing application layer and wireless network environment [8].
The application processing center is the nerve center of the intelligent warehouse
management system, which is responsible for connecting the ERP system and the mobile
terminal processing system [9]. The mobile terminal (PDA, PAD) obtains the source data

123
Design of Intelligent Warehouse Management System

All data is interactively

processed by the application
SAP Web Service
SAP database processing center
Interface

Field wireless Application

SAP database
network processing center

Web
application
Mobile Centralized
application printing Management
terminal applications query

Fig. 1 Application structure of intelligent warehouse management system

from the ERP database in real time through the ERPWeb Service interface in the wireless
network environment. In the intelligent warehouse management system, data processing
center is used for data collation. According to different business requirements, the intel-
ligent warehouse management system designs three subsystems, namely PDA subsystem,
PAD subsystem and PC subsystem [10].

3.2 Interface Mode

The third party warehouse management system interacts with the ERP system through the
interface, and there are four kinds of interface technology implementations. Write the
WMS temporary table, that is, ERP submits the document concurrently or delays writing
the WMS temporary table, this way provides temporary table by WMS. Tune WMS’s Web
Service, that is, ERP also submitted documents or delayed call WMS Web Service to
synchronize data to WMS. This method provides Web Service with WMS, and ERP
synchronously submits data to WMS by submitting or delaying the call of Web Service.
Read ERP temporary table, that is, temporary tables are provided by ERP. ERP writes
temporary tables of ERP while submitting documents. WMS accesses ERP temporary
tables at regular intervals, and synchronize data to WMS database. Tune ERP’s Web
Service, that is, ERP provides Web Service. When ERP submits a document, the data is
temporarily stored in the temporary table, and the WMS calls Web Service to query data
from the temporary table and synchronize the data to the WMS.
The above four interface methods have advantages and disadvantages: The temporary
table of the third party system can realize the real-time synchronization of data, but both
the ERP and the third party system do not stop accessing the temporary table, which easily
causes the resource shortage of the third party system server and leads to the operation
fault. At the same time, both systems in the temporary table work are difficult to distin-
guish the responsible party. The method of calling the third party system Web Service is
simple, and the ERP side development work is small. However, due to the influence of
network environment and the operation of the third party server, when the documents are
not submitted promptly, the real time data cannot be guaranteed. The way to read the
temporary table of ERP system has the advantages of simple implementation and small
workload of ERP side development. However, it is difficult to distinguish the responsibility
of the third party system when the system problem occurs. The way of adjusting ERP
system Web Service: the two systems are separated from each other, which is easy to

123
 J. Mao et al.

distinguish responsibility, but it is affected by ERP fault, server, network and other issues,
resulting in that ERP data cannot be timely synchronization to the risk of the third party
system. Interface technology differences are detailed in Tables 1 and 2.

3.3 Scheduling Algorithm Design

For scheduling algorithm design, we first initialize the population, set the population size
of popsize, and randomly generate the initial population. Unrestricted random population
may lead to infinite iterations of the algorithm, and it can’t converge. According to the
initial conditions given by the scheduling problem and the constraints that the solution of
the problem should satisfy, each scheme is initialized and repeated several times to obtain
the initialized population. The main constraint conditions of each scheme encoding m  p
matrix are the value of each column and quantity demanded of the material. The initial-
ization of each scheme follows the following algorithm:
First, we iterate the column number j to initialize the supply of each kind of material to
each warehouse. The variable tempBi temporarily records the amount of Kj material stored
in the warehouse Si, and its initial value is the Bij value in the formula (3–6). The variable
tempQ is the current demand of the Kj material, and the initial value is the value of Qj in
the formula (3 - 1), and the value will gradually decrease to 0. tempGi temporarily stores
the supply of Kj from storage Si. When tempQ [ 0, that is, the warehousing supply has not
yet met the demand, performs the random allocation operation. Randomly select a ware-
house Si, if its current material storage capacity of tempBi is greater than 0, it supplies part
of the material, and then it performs the next step, otherwise it returns and picks up the
warehouse. If tempQ = 0, execute the next step directly. TempBi [ 0, generates Si of
goods in warehouse temp_rand, but satisfies the constraints. At this point, the temp_rand is
accumulated into the tempGi, modifying the values in the tempQ, subtracting the tem-
p_rand. Then, we go back to step 2 to determine the value of tempQ. If tempQ = 0, the
demand for current supplies have been allocated. We return step 1 to initialize the supply
of the next material until all the goods are fully allocated to each warehouse and the
scheme is initialized.
The fitness function is used to evaluate the advantages and disadvantages of each
scheme in the population, which is the basis for the selection. In this scheduling problem,
the fitness function is designed based on the optimized target in the above. That is, the
storage resource utilization rate, the numbers of warehouses involved in scheduling and
scheduling time are considered to comprehensively evaluate the fitness of the program. The
fitness function f is shown in Eq. (1).
Pm  Pp xij 
i¼1 F i  j¼1 Bij
f ¼ P  12 ð1Þ
w1  11  m i¼1 Fi þ w2  maxfFi  ti g
xij
The parameter Fi is utilized to eliminate storage that not involved in scheduling. Bij
indicates the percentage of Kj material supplied by warehousing Si as a percentage of its
P x
supply. It reflects the degree of utilization of storage resources, and pj¼1 Bijij is the com-
prehensive evaluation of warehouse. All the participating warehouses are evaluated, and all
the material calls in the scheduling scheme are evaluated. Here, 12 corrects the results on
the whole. The higher the utilization ratio of warehouse resources, the higher the fitness,
the smaller the number of storage and scheduling time the denominator represents, the
higher the fitness. Therefore, the value of fitness function f is greater, indicating that the

123
 Table 1 Interface technology difference comparison
The interface way Impact on The coupling The real time The third party The ERP system is Impact on ERP Development
production of the two system is subject subject to the fault submission data tools of ERP
systems systems to the fault of the of the interface interface
interface
Design of Intelligent Warehouse Management System

Write a temporary list of third-party systems Higher High High Hard – Higher Bigger
Tune the third-party system Web Service The minimum Low High Easy – Lower The minimum
Read the ERP system temporary table Maxima High Low – Hard The minimum Smaller
Tune ERP system Web Service Lower Low Low – Easy Lower Bigger

123
 Table 2 Interface technology difference comparison
The way Advantage Disadvantages

123
Mode 1 way to write a Such as ERP submission of documents and at the same time the data is When the third-party system failure is difficult to distinguish between
third-party system written to a third-party system temporary table, enabling real-time responsibilities. As the temporary table is a shared table, ERP system
temporary table data synchronization needs to write data to a third-party system temporary table. In order to
ensure real-time synchronization of data, third-party systems need to
keep access to the temporary table. When the third-party system
server fails, you need to analyze the two programs to determine the
problem. However, in fact, the procedures of the two sides are
independent and the analysis of the problems is difficult to operate and
difficult to distinguish
Synchronously when the ERP submits the document, if the third party
system fails or the server or the network fail to call, the normal
submission of the documents on the ERP side will be affected
If the ERP side of the document does not immediately call the Web
Service, but through the timing of the delayed call Web Service
trigger, although the impact does not affect the ERP side of the
submission of documents. However, the data of the ERP can not be
synchronized with the third-party system in real time, which may
cause the third-party system to interrupt the business operation due to
the inability to obtain the proper data, thereby affecting the production
Due to changes in demand and other reasons need to change the
structure of the temporary table, the interface program needs to be
modified, will have additional development costs
ERP submission documents processing speed slower than before
J. Mao et al.
 Table 2 continued

The way Advantage Disadvantages

Mode 2 transfer third- To achieve a simple, ERP-side development workload. If the third party system Web Service is called synchronously when the
party system Web Such as ERP documents submitted at the same time call Web Service. ERP submits the document, if the third party system fails or the server
Service The two systems separated by the Web Service interface, mutual or the network fail to call, the normal submission of the documents on
influence, easy to distinguish between responsibilities. the ERP side will be affected
The design of a flexible Web Service interface to deal with future If the ERP side of the document does not immediately call the Web
changes in the interface changes caused by additional program Service, but through the timing of the delayed call Web Service
development trigger, although the impact does not affect the ERP side of the
submission of documents. However, the data of the ERP can not be
synchronized with the third-party system in real time, which may
cause the third-party system to interrupt the business operation due to
the inability to obtain the proper data, thereby affecting the production
ERP submission documents processing speed slower than before
Mode 3 read the To achieve a simple, ERP-side development workload. When the third-party system failure is difficult to distinguish between
temporary table of ERP documents submitted at the same time need to write data into the responsibilities. As the temporary table is a shared table, ERP system
ERP temporary table, but due to the temporary table in the ERP system needs to write data to a third-party system temporary table. In order to
Design of Intelligent Warehouse Management System

itself in the same system, so the speed has little effect ensure real-time synchronization of data, third-party systems need to
keep access to the temporary table. When the third-party system
server fails, you need to analyze the two programs to determine the
problem. However, in fact, the procedures of the two sides are
independent and the analysis of the problems is difficult to operate and
difficult to distinguish
If the ERP side of the document does not immediately call the Web
Service, but through the timing of the delayed call Web Service
trigger, although the impact does not affect the ERP side of the
submission of documents. However, the data of the ERP can not be
synchronized with the third-party system in real time, which may
cause the third-party system to interrupt the business operation due to
the inability to obtain the proper data, thereby affecting the production
Due to changes in requirements and other reasons need to change the
temporary table structure, the interface program needs to be modified,
will have additional development costs

123
 Table 2 continued

The way Advantage Disadvantages

123
Mode 4 tune ERP Web The two systems separated by the Web Service interface, mutual When the third-party system failure is difficult to distinguish between
Service influence, easy to distinguish between responsibilities. responsibilities. As the temporary table is a shared table, ERP system
The design of a flexible Web Service interface to deal with future needs to write data to a third-party system temporary table. In order to
changes in the interface changes caused by additional program ensure real-time synchronization of data, third-party systems need to
development. keep access to the temporary table. When the third-party system
Do not pass bi-directional log records to check the specific interface call server fails, you need to analyze the two programs to determine the
process problem. However, in fact, the procedures of the two sides are
independent and the analysis of the problems is difficult to operate and
difficult to distinguish
Call ERP Web Service, resulting in third-party system can not be real-
time access to ERP system data, it may lead to third-party systems due
to lack of access to the data and interrupt business operations, and thus
affect the production
J. Mao et al.
Design of Intelligent Warehouse Management System

fitness of the scheme is higher. The fitness function value fitvalue of each scheme in the
population is used as the basis for the selection in subsequent operations.
In view of the previous cloud storage logistics scheduling model, this section will apply
a bee colony genetic hybrid optimization scheduling algorithm to optimize the solution and
search for the best solution. Aiming at the specific problems, the author designs the
selection operator, crossover operator and mutation operator in the algorithm, and gets the
best fitness solution after generations of propagation optimization.

4 Result Analysis and Discussion

4.1 System Function Test

Logistics management module achieves the logistics tracking, logistics information man-
agement, logistics management. The functions of the three parts are tested, and the per-
formance of the hybrid genetic algorithm is simulated and analyzed. For logistics tracking,
delivery list of specific odd number is located and queried by GPS. The electronic map
matches its current location, and displays the outgoing number, vehicle number plate,
location, shipping warehouse and other information. For the logistics information man-
agement part, it displays the detailed list in the management scope. When click any specific
outbound list, you will enter the specific operation page. In the logistics information
management detailed operation page, it displays the basic information and detailed list of
materials. The shipper will confirm the quantity of each item by manual input of actual
quantity and material confirmation, and then click to confirm receipt after confirmation by
item. At this point, the system will make a corresponding judgment, if the number of the
library and the actual number of receiving does not match, the document needs to be
further processed, and the logistics information management list will continue to retain the
document. If the number of the displayed quantities is consistent with the actual number of
received goods, the receipt will be successful, and the document will be removed from the
logistics information management list. Due to the limited space, we will not repeat the test
of other items.
The logistics scheduling management module uses the bee colony genetic hybrid
optimization scheduling algorithm described in the previous paper. Firstly, the simulation
of the optimal scheduling algorithm is carried out, and its performance is analyzed. The
initial conditions of the algorithm are set up, in which the storage quantity is m = 6, the
demand material type is p = 5, the storage and storage time of each warehouse and the
delivery time are shown in Table 3. The material requirements for scheduling are shown in
Table 4. Population size is popsize = 100, genetic algebra is limited to num-genera-
tion = 50, cross probability is Pc = 0.6, variation probability is Pm = 0.001. In the fitness
function, w1 = 0.6, w2 = 0.4, n1 = 10, n1 = 100.
The simulation analysis of performance optimization scheduling algorithm for mixed bee
swarm genetic below is proposed in this paper. In the initial conditions, it performs optimal
scheduling algorithm. The fitness function values of each generation scheme in the genetic
process are recorded, and the maximum fitness value, the average fitness value of the top 10
and the mean value of fitness values for all schemes are marked. As shown in Fig. 2.
The abscissa in the graph represents the algebra of heredity, and the maximum value of
the fitness in the population is marked by the vertical starting point from the curve to the
height (marked by number 1). The mean value of the first 10 maximum fitness values in the

123
 J. Mao et al.

Table 3 Warehouse basic data

Material types Delivery time

K1 K2 K3 K4 K5 T

Warehousing S1 1300 200 60 3000 1200 20

Warehousing S2 1500 310 40 2000 1600 30
Warehousing S3 600 200 30 1000 600 50
Warehousing S4 800 500 20 1800 500 25
Warehousing S5 900 150 50 800 1400 40
Warehousing S6 2200 200 10 200 800 35

Table 4 Material demand data

Material types

K1 K2 K3 K4 K5

Quantity demanded 2000 300 50 4200 1600

population is marked as number 2, and the average fitness value of all the alternatives in
the population is marked as number 3. According to the graph analysis, the starting points
of the three curves are different. Especially, the starting points of curve 1 and curve 3 are
quite different, which shows that the initialization population generated by the algorithm
has certain randomness. With the running algorithm of generations, the difference between
the three curves is getting smaller, and all of them are on the rise, and eventually converge
to the same value, which shows that the algorithm does optimize the scheme of the
population in the process of operation, which has a certain convergence.
The algorithm is run 100 times under the above initial conditions, and the optimal
fitness value of each program is recorded, and the curve is plotted, as shown in Fig. 3.
According to the graph, the best fitness values obtained by the running algorithm are
slightly different under the same initial conditions, which are related to the setting of some
random parameters in the algorithm. But the running results of the best fitness value are

Fig. 2 The fitness of each generation of the algorithm and its mean curve were compared

123
Design of Intelligent Warehouse Management System

concentrated in the range of 2226, the fluctuation range is not large, and the algorithm has
certain stability.
The optimal scheduling algorithm combines the domain search in bee colony algorithm
with the improved genetic algorithm. In this paper, the final algorithm is compared with the
original genetic algorithm, the improved genetic algorithm, and the bee colony algorithm.
With the same initial conditions mentioned in the front, through the fitness performance
curve of original genetic algorithm, improved genetic algorithm, bee colony algorithm, and
bee colony genetic hybrid optimization scheduling algorithm with neighborhood search,
we can know that the convergence of genetic algorithm is relatively fast. Bee algorithm is
slower than the genetic algorithm, but it can get better solution. Compared with the other
three algorithms, the optimal scheduling algorithm proposed in this paper has a slower
overall convergence, but it can find the relatively optimal solution. The four algorithms are
run 100 times under the same initial conditions, and the optimal fitness values are obtained
at each run. The minimum, mean, and maximum values of the best fitness value of the
corresponding algorithm in the 100 run are calculated, and the results are plotted. In the
100 run, the minimum fitness, the average value and the maximum fitness of the optimal
scheduling algorithm proposed in this paper are better than the other three algorithms.
Under different population size popsize, the four algorithms are compared with each
other. The population size is set to 20, 50 and 100 respectively. The genetic algorithm, the
improved genetic algorithm, the bee colony algorithm, and the optimized scheduling
algorithm curve are proposed. We can conclude that the larger the population size, the
better the fitness that can be reached. This is because the larger the population size, the
more alternatives. So the program is more diverse, which is easier to get better programs.
At the same time, the optimal scheduling algorithm proposed in this paper can be better
than the other three algorithms in different population size. The bee colony genetic hybrid
optimization scheduling algorithm is applied to the logistics scheduling management
module of the system. In the system, there is a list of statistical requirements for each
requirement. Enter the specified demand odd number into the system, and click the dis-
patch allocation button, the system calls the optimization algorithm to search data in the
related data table. The schedule of the requirements list is feedback by the page.

Fig. 3 The optimal fitness curve of the algorithm is given

123
 J. Mao et al.

In this chapter, the main functional modules of the system, namely warehouse management
module and logistics management module and its sub modules, are run and tested to verify the
functions provided by each part. The simulation of bee colony genetic hybrid optimization
scheduling algorithm is carried out, and it is proved that although the convergence is slow, it is
beneficial to get better scheduling solution, which has certain robustness.

5 Conclusion

One of the most important things in logistics operations is the transit of goods from finished
goods to use. Since it is a transit, there must be a storage process in the middle, which involves
the management of materials in the storage, that is, the problem of warehouse management. In
this paper, the function framework of intelligent warehouse management system based on
cloud model is designed, and a hybrid genetic algorithm based on bee colony is proposed to
solve the problem of cloud storage logistics scheduling. The system makes use of the RFID
technology and GPS technology to obtain real-time data. The intelligent warehouse man-
agement system is built to provide users with the two main parts of warehouse management
and logistics management services, and realize the intelligent monitoring and management of
the entire storage life cycle of materials. In this paper, the cloud storage logistics resource
scheduling problem is modeled under the actual environment, and the evaluation index system
of logistics resource scheduling is constructed. Combined with the real time perception
information of warehouse and the key factors affecting logistics scheduling, this paper studies
the scheduling evaluation and decision-making of logistics resources in the cloud storage
environment, and proposes an optimal scheduling method. Finally, we run and test the main
functional modules of the system, and make simulation analysis. The test results show that
although the convergence of the algorithm in the system is slow, but it is conducive to get
better scheduling solution, which has certain robustness.

Funding Funding was provided by China Scholarship Fund and Jilin provincial science and technology
department project in 2017 (Grant No. 20170418058FG).

References
1. Yang, L. Q., & Bi, Y. Y. (2015). Internet of things technology implementation by applying SDLC
model: The intelligent storage management system. Applied Mechanics and Materials, 556–562(3),
5385–5390.
2. Zhang, Y., Liu, B., Zhang, T., et al. (2015). An intelligent control strategy of battery energy storage
system for microgrid energy management under forecast uncertainties. International Journal of Elec-
trochemical Science, 9(8), 4190–4204.
3. Sami, B. S., & Jeddah, S. (2017). An intelligent power management investigation for stand-alone hybrid
system using short-time energy storage. International Journal of Power Electronics & Drive Systems,
8(1), 10.
4. Fleiner, C., Garner, R. B., Hafner, J. L., et al. (2016). Reliability of modular mesh-connected intelligent
storage brick systems. Ibm Journal of Research & Development, 50(2.3), 199–208.
5. Yoo, C. H., Chung, I. Y., Lee, H. J., et al. (2016). Intelligent control of battery energy storage for multi-
agent based microgrid energy management. Energies, 6(10), 4956–4979.
6. Yan, Q., Xing, C. X., & Zhou, L. Z. (2017). Design and implementation of the storage node man-
agement system in CNGI-MSN. Journal of Chinese Computer Systems, 28(8), 1532–1536.
7. Chauhan, R. K., Rajpurohit, B. S., Gonzalez-Longatt, F. M., et al. (2016). Intelligent energy man-
agement system for PV-battery-based microgrids in future DC homes. International Journal of
Emerging Electric Power Systems, 17(3), 339–350.

123
Design of Intelligent Warehouse Management System

8. Bottani, E., Montanari, R., Rinaldi, M., et al. (2015). Intelligent algorithms for warehouse management.
Intelligent Systems Reference Library, 87(1), 645–667.
9. Duan, S., & Xu, Y. (2016). Construction of intelligent logistics warehouse management information
system based on RFID technology. Mining & Processing Equipment, 38(21), 7410–7421.
10. Wang, S. H., Information, S. O., & University, B. W. (2016). Application of RFID in heavy truck
logistics warehouse management. Logistics Sci-Tech, 6(1), 128–137.

Jia Mao male, born in 1973, is an associate professor in College of

Transportation, Jilin University. His research direction is logistics
system simulation.

Huihui Xing female, born in 1991, is a postgraduate student in College

of Transportation, Jilin University. Her research direction is logistics
system simulation.

Xiuzhi Zhang female, born in 1973, is an associate professor in

Institute of Mechanical Science and Engineering, Jilin University. Her
research direction is mechanical engineering.

123