You are on page 1of 14

Applied Soft Computing Journal 91 (2020) 106207

Contents lists available at ScienceDirect

Applied Soft Computing Journal


journal homepage: www.elsevier.com/locate/asoc

A novel hybrid MCDM model for machine tool selection using fuzzy
DEMATEL, entropy weighting and later defuzzification VIKOR

Hai Li a , Wei Wang a , , Lei Fan a , Qingzhao Li a , Xuezhen Chen b
a
School of Mechanical and Electrical Engineering, University of Electronic Science and Technology of China, Chengdu, 611731, China
b
Department of NC Processing Plant, AVIC Chengdu Aircraft Industrial (Group) Co., Ltd., Chengdu, 610073, China

article info a b s t r a c t

Article history: Machine tool selection has been an important issue in the manufacturing industry because improper
Received 16 July 2019 machine tool selection can have a negative effect on productivity, accuracy, flexibility, and the re-
Received in revised form 17 February 2020 sponsive manufacturing capabilities of a company. The current multi-criteria decision making (MCDM)
Accepted 2 March 2020
approach of machine tool selection mostly focuses on the subjective perspective. However, as the
Available online 5 March 2020
objective evaluation represents the actual performance of machine tools, both subjective and objective
Keywords: perspectives need to be considered when choosing an appropriate machining tool. Therefore, this study
Machine tool selection proposes a machine tool selection method based on a novel hybrid MCDM model. Firstly, the presented
Hybrid MCDM model method employs a comprehensive weight technique integrating subjective weights obtained using
Fuzzy DEMATEL fuzzy decision-making trial and evaluation laboratory (FDEMATEL) with objective weights obtained
Entropy weighting
using entropy weighting (EW). Secondly, later defuzzification VIKOR (LDVIKOR) is put forward to rank
Later defuzzification VIKOR
the optional alternatives. Finally, a case application verifies the effectiveness of the proposed method.
The evaluation results indicate that the best and worst selected machine tool of the proposed method
keeps high conformance with the actual ranking in real factory. Additionally, sensitivity analysis results
of the effect of parameters ϕ on the decision outcome show that irrespective of the variations in this
parameter, the best decision outcome will be not influenced. These indicate that the presented hybrid
model has advantages in granting flexibility to the preferences of decision makers.
© 2020 Published by Elsevier B.V.

1. Introduction the flexibility and quality of products, and enhance the overall
productivity [3].
Manufacturing industries are facing multiple challenges such On the contrary, if the machine tool is not selected properly,
as globalization and rapidly changing market demand, result in it will cause damage to the machine tool. Over 300 companies
higher manufacturing requirements and more complex parts [1]. in the U.S can provide machine tool remanufacturing services.
Therefore, it is critical for the profitability of enterprises to se- It has been proved that the cost of machine tool downtime and
lect appropriate machining equipment for specific manufacturing maintenance due to failure or improper use is close to 75% of
tasks. As an efficient, economic, high value-added and complex the cost of new machine tools [4]. There are more than 2000
structure electromechanical equipment, machine tool provides enterprises in China that can provide maintenance and reman-
key manufacturing capacity to turn raw materials into finished ufacturing services for machine tools [5]. The research shows
parts for final product assembly, which is mainly used for large- that, however, the main reason for machine tool downtime and
maintenance is still due to failure or improper use of machine
scale mechanical equipment of ships, construction machinery,
tools. Therefore, in the face of specific processing tasks, how
railway vehicle, wind turbine, hydroelectric generator, nuclear
to select the proper machine tool from various candidates of
power equipment, petrochemical equipment, and so on [2]. For
machine tools has always been a major difficulty for enterprise
specific machining tasks, manufacturing enterprises usually have
decision makers, as the improper selection of machine tools may
multiple machine tools that can meet the machining require-
could negatively affect the overall performance of manufacturing
ments, and need to consider the cost, productivity and enterprise
system [6].
profit to select appropriate machine tools. Choosing the correct
Machine tool selection is usually a difficult task for engineers
machine tool can decrease the production delivery time, improve
and managers, as there are many qualitative and quantitative
attributes that need to be considered in the selection of the
∗ Corresponding author. appropriate machine tool [7]. Many efforts have been devoted to
E-mail address: wangwhit@163.com (W. Wang). performing machine tool selection. The current methods mainly

https://doi.org/10.1016/j.asoc.2020.106207
1568-4946/© 2020 Published by Elsevier B.V.
2 H. Li, W. Wang, L. Fan et al. / Applied Soft Computing Journal 91 (2020) 106207

include MCDM model and optimization-based model. The optimal tool selection and analyzes the decision results. The implication
mathematical model can deal with the objective data well, but it of study includes managerial implication and social implication is
often ignores the qualitative and subjective considerations. Due presented in Section 6. Conclusions and future scope are afforded
to the diversity of conflicting machine tool selection criteria, the in Section 7.
quantification of uncertain qualitative attribute information can
be extremely complex. For example, a machine tool that is able 2. Related review
to manufacture several different parts could result in different
auxiliary time, clamping time, machining time and changeover 2.1. Machine tool selection evaluation criteria
time [8]. Therefore, an effective machine tool selection needs to
compromise over conflicting tangible and intangible factors, and Establishing evaluation criteria should be the top priority in
to address these issues, multiple criteria decision making (MCDM) the machine tool selection process. In previous studies, machine
has been found to be useful. However, expert assessment of tool selection utilized various criteria. Cagdac et al. [12] applied
qualitative attributes is subjective and therefore imprecise due to ten criteria (flexibility, precision, safety, space, productivity, re-
the decision-makers and uncertainty [9]. As the evaluation infor- liability, cost, environment, adaptability, and maintenance) to
mation such as criteria weights and alternative ratings are usually select machine tools. Cimren et al. [13] selected criteria such
expressed in linguistic terms, the fuzzy linguistic approach has as environment, flexibility, productivity, safety, and adaptability.
been found to be a suitable method to handle the problem of Samvedi et al. [14] employed seven criteria, including cost, safety,
expert assessment. installation easiness, compatibility, operative flexibility, produc-
However, the MCDM model used in machine tool selection tivity, and user-friendliness. Ayag et al. [15] selected flexibility,
mainly focuses on the combination of fuzzy Analytic Hierarchy productivity, adaptability, maintenance and service, space, pre-
Process (AHP) and other MCDM methods, ignoring the interde- cision, safety and environmental, and reliability to evaluate ma-
pendence between the criteria. Besides, machine tool is a high chine tool alternatives. Nguyen et al. [16] considered five criteria,
value-added equipment. To improve product quality and produc- specifically, including damping capacity, stiffness, speed capacity,
tivity, the interdependence between decision criteria must be thermal stability, and accuracy. Aghdaie et al. [17] adopted cost,
considered to fully exert the potential of machine tools. Addi- operative flexibility, maintainability, serviceability, size, physi-
tionally, the current selection method of MCDM is mainly based cal, and productivity to choose machine tools. From the above
on the expert experience judgment. Actually, as service time analysis, the criteria of cost, productivity, reliability, safety, and
increases, most of the machine tools have potential problem flexibility were most commonly employed in selecting machine
such as deterioration of accuracy and change of working con- tools. As for a specific machining task, we should consider the ac-
dition, resulting in the dynamic change of machine tool perfor- tual performance like failure rate, utilization and linkage accuracy
mance [10,11]. Therefore, only considering the subjective evalu- to comprehensively evaluate machine tool.
ation method, the performance of the selected machine tool by
MCDM model may be difficult to satisfy the actual machining 2.2. Machine tool selection approach
requirements, even may damage machine tool.
From the existing researches, no suitable decision-making Machine tool selection mainly contains two types includ-
methods are mentioned to select the reasonable alternative of ing optimization-based model and criteria-based method. Some
the machine tool from both subjective and objective perspectives. studies proposed optimization-based methods for machine tool
To fill this gap that the current decision-making problem for selection problems involving continuous variables. Rai et al. [18]
machine tool selection ignores the interdependence between the proposed a machine tool selection model with constraints (such
criteria and actual operation information, this study intends to as capacity, tool life, and magazine) to minimize the total cost,
an establish decision-making model for machine tool selection and this model was optimized using a genetic algorithm. Mishra
from the perspective of combining expert knowledge and actual et al. [19] developed a fuzzy goal-programming model to ad-
performance of machine tool. dress the machine tool selection problem, and quick converging
The novelty in this study is to develop a machine tool se- simulated annealing was employed to solve this model. Jahromi
lection approach considering both expert knowledge and actual et al. [20] presented novel zero-one linear integer program-
performance of alternatives based on proposed fuzzy DEMATEL- ming policies to select the machine tools dynamically. Perhin
EW-LDVIKOR model, making the decision-making results closer and Min [21] developed a hybrid decision-aid tool combining
to the actual situation and getting more reliable and practical fuzzy linear regression and quality function deployment zero-
evaluation results. one goal programming to obtain the decision results of machine
One of the main contributions in this study is to develop tool selection. He et al. [22] suggested a machine tool selection
a machine tool selection method using both expert knowledge method based on an energy-saving optimization model and the
and actual operation information to provide the optimal machine nested partitions framework. Liu et al. [23] presented an efficient
tool for a specific manufacturing task. Another contribution is machine tool selection method based on energy efficiency eval-
to develop a hybrid MCDM model named fuzzy DEMATEL-EW- uation using the modeling machine tool-related factors of each
LDVIKOR that enables the alternative evaluation from subjective alternative and workpiece-related factor of expected tasks.
and objective perspectives and provides the best alternative for In addition to the optimization-based approaches, the criteria-
decision makers. based model have been widely utilized to address the machine
This paper aims to proposed a decision-making method for tool selection problem, like AHP [24], VIKOR [25], Technique for
solving the problem of alternative selection in machine tool, Order Performance by Similarity to Ideal Solution (TOPSIS) [26],
which can avoid the subjective of traditional decision-making Preference Ranking Organization Method for Enrichment Evalu-
method based on experience. The paper is organized as follows. ation (PROMETHEE) [27], Analytical Network Process (ANP) [28],
A literature review will be provided to summarized and analyze Stepwise Weight Assessment Ratio Analysis (SWARA) [29], Com-
relevant research works in Section 2. In Section 3, we will provide plex Proportional Assessment (COPRAS) [30]. Among the criteria-
the basic theory of fuzzy set and fuzzy operation. Section 4 based approaches, the AHP technique has been most popular
proposes a hybrid decision-making method based on DEMATEL, as it can handle both quantitative and qualitative data. Some
EW and LDVIKOR. Section 5 describes the case study of machine hybrid methods based on AHP have also been proposed. Önüt
H. Li, W. Wang, L. Fan et al. / Applied Soft Computing Journal 91 (2020) 106207 3

et al. [7] employed the TOPSIS to evaluate and select vertical Table 1
machining centers and fuzzy AHP for calculation of the criteria Criteria-based MCDM approaches in machine tool selection.

weight. Dagdeviren [31] proposed a hybrid method using the AHP MCDM technique Reference source
and preference ranking organization method for enrichment eval- TOPSIS and fuzzy AHP Önüt et al. [7]
uation (PROMETHEE) to solve the equipment selection problem. Fuzzy AHP and Gray relational analysis Samvedi et al. [14]
Fuzzy ANP and TOPSIS Ayağ and Özdemir [15]
Yurdakul and Ic [32] utilized the TOPSIS using fuzzy numbers for SWARA and COPRAS Aghdaie et al. [17]
machine tool selection. Özgen et al. [33] combined AHP, modified AHP and PROMETHEE Dagdeviren et al. [31]
Delphi, and PROMETHEE to perform machine selection. Taha and Fuzzy TOPSIS Yurdakul and Ic [32]
Rostam [34] developed a method of integrating PROMETHEE and AHP and DELPHI & PROMETHEE Özgen et al. [33]
AHP to select the most suitable machine. Samvedi et al. [14] inte- Fuzzy AHP and PROMETHEE Taha et al. [34]
Fuzzy AHP and FVIKOR Ilangkumaran et al. [35]
grated Gray relational analysis and fuzzy AHP to select the most FANP and ANN Sadeghian and Sadeghian [36]
appropriate machine tool. Ilangkumaran et al. [35] presented
a hybrid fuzzy AHP and fuzzy VIKOR model for the selection
of an optimal machine tool among various alternatives. Ayağ
and Özdemir [15] solved the problem of machine tool selec- given increasing attention in recent years. The findings of the lit-
tion by utilizing ANP and TOPSIS. Aghdaie et al. [17] employed erature review can be summarized as follows. First, most studies
SWARA and COPRAS in decision making in machine tool selection. in machine tool selection only focused on subjective evalua-
Also, Sadeghian and Sadeghian [36] integrated artificial neural tion, which has given rise to weakened practicality, and thus,
networks (ANNs) and fuzzy ANP to select machine tools. Some re- it is difficult to promote them in real manufacturing factories.
searches employed the other MCDM approaches to determine the Second, the MCDM methods are widely used than optimization-
assessment values of the criteria. Table 1 lists the criteria-based based methods in machine tool selection. However, most pre-
strategies for machine tool selection. vious studies employed fuzzy AHP with other MCDM methods,
and these hybrid methods have limitations in relatively complex
2.3. Hybrid MCDM models
calculations and criteria dependencies. Based on the above find-
ings, this research proposes a hybrid method termed FDEMATEL-
There are some techniques have been developed to address
EW-LDVIKOR to solve the problem of machine tool selection,
MCDM problems, such as AHP, VIKOR, COPRAS, SWARA, DE-
considering subjective and objective evaluations (Table 3).
MATEL [37], Best–Worst method (BWM) [38], Full Consistency
method (FUCOM) [39], ANP, Data Envelopment Analysis (DEA)
[40], TOPSIS, Elimination and Choice Expressing Reality (ELECTRE) 3. Mathematical preliminary
[41], Strengths Weaknesses Opportunities and Threats (SWOT)
[42], Multi-Attribute Ideal Real Comparative Analysis (MAIRCA) Since the proposed hybrid methodology integrates the fuzzy
[43], Combinative Distance-based Assessment (CODAS) [44], etc.
set theory for manipulating information uncertainty, this section
The combination of these approaches is called hybrid MCDM
introduces some essential characteristics of fuzzy set theory and
technique, and it shows the apparent advantages and strong
fuzzy operations.
robustness, like DEMATEL-ANP-VIKOR [45], FUCOM-COPRAS [46],
SWOT-AHP [47], and DEA-TOPSIS-COPRAS [48]. Therefore, it has
gained many applications in different fields (Table 1). The tra- 3.1. Fuzzy set theory
ditional crisp MCDM model has some shortcomings, such as
it cannot handle the linguistic judgment vagueness and group Fuzzy set theory is a mathematical theory first introduced
preference diversity. Thus, some researches have explored the
by Zadeh [74], which is designed to model the vagueness or
fuzzy-based technique to investigate the qualitative information
imprecision of human cognitive processes. The basic idea of fuzzy
of alternatives, particularly for decision-making problems and
set theory is that the elements have a degree of membership
intelligent evaluation issues [49]. Recently, as the fuzzy-based
in a fuzzy set. Fuzzy set is defined by a membership function
theory can describe uncertain and incomplete information, like
fuzzy set theory [50], rough set theory [51] and neutrosophic [52], that maps elements to degrees of membership within a certain
some researchers added it into MCDM approaches to establish interval, which is usually [0,1]. Different types of fuzzy mem-
fuzzy MCDM models, such as fuzzy DEMATEL (FDEMATEL) [53], bership functions have been used in fuzzy logic. There are three
rough DEMATEL [54], rough-fuzzy DEMATEL [55], graphics-based most common types, including monotone, triangle and trape-
rough fuzzy DEMATEL [56], two-dimension uncertain linguis- zoidal. As fuzzy set is convex function, trapezoid function or
tic [57], fuzzy VIKOR [58], Pythagorean fuzzy VIKOR [59], and trapezoidal function approach the convex function the best [75].
neutrosophic-based DEMATEL [60]. Similarly, the combination of However, triangular fuzzy number (TFN) is more convenient in
fuzzy set theory and these MCDM models is termed the hybrid application due to its simple calculation [76], and it is useful in
fuzzy MCDM models, like FDEMATEL and TOPSIS [61], FDEMATEL promoting representation and information processing in fuzzy
and fuzzy ANP [62], rough DEMATEL, ANP and MAIRCA [63], environment [77]. Besides, TFN is the most suitable property
FVIKOR and ELECTRE [64], fuzzy AHP and FVIKOR [65], and fuzzy for experts’ linguistic evaluation, and it is widely utilized in
ANP and FVIKOR [66]. These models have been widely utilized in fuzzy MCDM researches [78]. The membership degree function
many practices, as shown in Table 1. Additionally, there are some of a triangular fuzzy number g̃ = (g l , g m , g u ) can be expressed
combinations of other MCDM methods under fuzzy environment, mathematically as follows:
such as fuzzy TOPSIS and AHP [67], fuzzy TOPSIS and SWOT- ⎧
AHP [68], fuzzy AHP and fuzzy MAIRCA [69], fuzzy Best–Worst, ⎨(x − g l )/(g m − g l ) g l ≤ x ≤ g m
relative entropy and VIKOR [70], fuzzy SWARA and Weighted µg̃ (x) = (g u − x)/(g u − g m ) g m ≤ x ≤ g u (1)
Aggregated Sum Product Assessment (WASPAS) [71], fuzzy AHP x < g l or x > g u
⎩0
and COPRAS [72], and Fuzzy BWM and CODAS [73]. Table 2 shows
their applications in different fields. where the g l , g m, and g u are the lower limit, middle limit, and
From the literature review on hybrid MCDM models and ma- upper limit of the triangular fuzzy number g̃, respectively. They
chine tool selection, it can be stated that these topics have been satisfy g l ≤ g m ≤ g u .
4 H. Li, W. Wang, L. Fan et al. / Applied Soft Computing Journal 91 (2020) 106207

Table 2
Recent applications using hybrid MCDM models.
Hybrid approach Specific applications
DEMATEL, ANP and VIKOR [45] Advertising media evaluation
FUCOM and COPRAS [46] Select sustainable supplier
SWOT and AHP [47] Identify specific factors of establishing plantations
DEA, TOPSIS and COPRAS [48] Select material
FDEMATEL and TOPSIS [61] Estimate participants
FDEMATEL and fuzzy ANP [62] Construct critical success factors
Rough DEMATEL, ANP and MAIRCA [63] Select bidder in the state administration
FVIKOR and ELECTRE [64] Solve outranking problems
Fuzzy AHP and FVIKOR [65] Evaluate sectoral investments
Fuzzy ANP and FVIKOR [66] Select site
Fuzzy TOPSIS and AHP [67] Evaluate and select mobile health applications
Fuzzy TOPSIS, SWOT and AHP [68] Evaluate the strategies for sustainable energy planning
Fuzzy AHP and fuzzy MAIRCA [69] Analyze failure modes and effects
Fuzzy Best–Worst, relative entropy and VIKOR [70] Analyze failure modes and effects
Fuzzy SWARA and WASPAS [71] Select shopping malls
Fuzzy AHP and COPRAS [72] Select market segment
Fuzzy BWM and CODAS [73] Select a site for a construction project

Table 3
Comparison between the extant researches and the proposed scheme.
Scope Extant research Proposed scheme
Machine tool selection Most hybrid approaches mostly focused on subjective The proposed scheme develops a machine tool selection
evolution without considering the actual performance data approach, considering both criteria interdependence and actual
[45–47,61,62]. performance data of all alternatives.
Hybrid MCDM model A. There are some hybrid methods based on the combination The proposed scheme presents a hybrid model
of DEMATEL, VIKOR, ANP, AHP, FUCOM [45–47,61,62]. FDEMATEL-EW-LDVIKOR from subjective and objective
B. There are some hybrid methods based on the combination perspectives and the preference of decision-makers.
of DEMATEL, VIKOR, ANP, AHP, FUCOM [45–48,61–63].

3.2. Fuzzy operation aggregating the evaluation information of experts in Eq. (10).
K K K
Suppose there are two TFNs à = (al , am ,
∑ ∑ ∑
xlij = λk xlkij , xm
ij = λk x m
kij , xij =
u
λk xukij (10)
a ) and B̃ = (b , b , b ), where a , a , a , b , b , b >0. Some fuzzy
u l m u l m u l m u
k=1 k=1 k=1
operations are presented as follows
Addition: where K ∑ denotes the total number of decision-makers and λk
K
1 λk = 1 where λ = (λ1 , λ2 , . . . , λk ) and λk ≥ 0.xij ,
l
satisfies
à ⊕ B̃ = (al , am , au ) ⊕ (bl , bm , bu ) = (al + bl , am + bm , au + bu ) (2) xm and x u
denote the lower limit, middle limit and upper limit of
ij ij
Subtraction: the evaluation information, respectively.
The defuzzification step is particularly crucial to defuzzify
à ⊙ B̃ = (al , am , au ) ⊙ (bl , bm , bu ) = (al − bl , am − bm , au − bu ) (3) TFNs into a crisp value. Previous studies utilized various defuzzi-
fication techniques such as the centroid technique [79], graded
Multiplication:
mean integration representation (GMIR) [80], and integral di-
à ⊗ B̃ = (al , am , au ) ⊗ (bl , bm , bu ) = (al × bl , am × bm , au × bu ) (4) vision [81]. In this study, GMIR, which has the advantage of
avoiding a zero in the denominator, is used to formulate the
Division: defuzzification, as in Eq. (11).
à = (al , am , au )−1 = (1/al , 1/am , 1/au )
{ −1
(5) xlij + 4 × xm u
ij + xij
B̃−1 = (bl , bm , bu )−1 = (1/bl , 1/bm , 1/bu ) xdef
ij = (11)
6
Maximum:
4. The proposed hybrid MCDM model
MAX (Ã, B̃) = (max(al , bl ), max(am , bm ), max(au , bu )) (6)
4.1. Fuzzy DEMATEL approach for subjective weights
Minimum:
MIN(Ã, B̃) = (min(al , bl ), min(am , bm ), min(au , bu )) (7) The Geneva Research Centre of the Battelle Memorial Institute
originally developed the DEMATEL technique to visualize the
Sum of product: structure of complicated causal relationships, which can help
n
∑ n
∑ DMs to more easily understand the interdependence of criteria
⊕wi ⊗ ãi = ⊕(w1 , w2 , . . . , wn ) ⊗ (ã1 , ã2 , . . . , ãn ) through matrixes or diagraphs. In the DEMATEL technique, DMs
i=1 j=1 are required to specify the relative importance and influence of
each criterion, which is challenging. Therefore, to address the
= w1 ã1 ⊕ w2 ã2 ⊕ · · · wn ãn (8)
vagueness, uncertainty, and information leakage, in this study,
Multiplication by a scalar: the fuzzy set theory is incorporated into the DEMATEL technique
in a hybrid approach termed FDEMATEL to obtain subjective
λa , λam , λau , if λ ≥ 0
{( l )
λ ⊗ Ã = ( u (9) weights, which contains five steps.
λa , λam , λal , if λ < 0
)
Step 1: Construct a fuzzy initial direct-influence matrix T̃ .
Suppose that the weight of kth decision-maker is λk , the Let C = {C1 , C2 ,. . . , Cn } be a set of criteria, where Cj denotes
initial direct-influence average fuzzy matrix can be obtained by the jth criterion, j = 1, 2, . . . , n. n is the total number of criteria.
H. Li, W. Wang, L. Fan et al. / Applied Soft Computing Journal 91 (2020) 106207 5

Table 4 4.2. Entropy weighting technique for objective weights


Defined linguistic variables for evaluation information.
Linguistic variables Abbreviations TFN Entropy is a concept in thermodynamics, first introduced to
No influence NI (0,0,0.25) information theory by C.E. Shannon, which is called informa-
Very low influence VL (0,0.25,0.5) tion entropy [83]. Entropy weight method is an objective weight
Low influence L (0.25,0.5,0.75)
High influence HI (0.5,0.75,1)
method. During specific use, it can achieve a relative objective
Very high influence VH (0.75,1,1) attribute weight by first using information entropy to calculate
the entropy weight of each attribute and second using entropy
weight to revise the weight of each attribute according to the
degree of variation of each attribute. Additionally, the entropy
Let E = {E1 , E2 ,. . . , Ek } be a set of experts, where Ek denotes method is especially valuable to examine disparities between sets
the kth expert, k = 1,2,. . . , K. K is the total number of experts. of information, and if the same values obtain for certain attributes
l
tkij = (tkij , tkij
m
, tkij
u
) is the initial direct-influence information that for different alternatives, the attributes need to eliminate [84].
l m
Ci has on Cj by kth expert using linguistic variables, where tkij , tkij , Now, it has been widely applied in the fields of engineering
u
tkij denote the lower limit, middle limit, upper limit of the ini- technology and social economics [85]. The steps for obtaining the
tial direct-influence information, respectively. Table 4 lists these objective weight W obj can be summarized as follows:
linguistic variables for fuzzy direct-influence.
Step 1: Normalize the actual performance data.
T̃ij = (t1ij ⊕ t2ij ⊕ · · · ⊕ tkij )/k (12) xij
vij = ∑m (17)
i=1 xij
Step 2: Compute the normalized direct-influence matrix Ỹ by
where vij denotes the normalized value of alternative Ai about Cj .
normalizing T̃ . xij means the crisp value of alternative Ai with respect to Cj . m
Ỹij = T̃ij /r = (tijl /r , tijm /r , tiju /r) = (ylij , ym represents the total number of evaluated alternatives.
ij , yij )
u
(13)
Step 2: Calculate the entropy value of jth criterion.
where Ỹij is the normalized direct-influence information that Ci
m m
has on Cj , where ylij , ym u
ij , yij denote the lower limit, middle limit,
∑ 1 ∑
upper limit of the normalized direct-influence matrix, respec-
zj = −k vij In(vij ) = − vij In(vij ) (18)
∑n In(m)
i=1 i=1
tively. r is MAX1≤i≤n ( j=1 tiju ) and MAXi means the maximum of
ith criterion. where In(•) is logarithm based on e and zj is [0, 1].

Step 3: Obtain the total-influence matrix P̃. Step 3: Compute the objective weight of jth criterion.
1 − zj
P̃ = lim (Ỹ 1 ⊕ Ỹ 2 ⊕ · · · ⊕ Ỹ k ) = Ỹ × (1 − Ỹ )−1 (14) Wj
obj
= ∑n (19)
k→∞
j=1 (1 − zj )
where P̃ij = (plij , pm ij , pij ) and P̃ij is the total-influence information
u
4.3. Comprehensive weight technique
that Ci has on Cj , where plij , pm u
ij , pij denote the lower limit, middle
limit, upper limit of the total-influence matrix, respectively. Here To describe the preference of each DM, we introduce the
plij = ylij × (1 − ylij )−1 , pm m m −1
ij = yij × (1 − yij ) and puij = yuij × (1 − yuij )−1 . weight preference factor ϕ to generate the comprehensive weight
Step 4: Establish the influential relation map (IRM). Wjcom of jth criterion by integrating the objective and subjective
In this step, the sum of rows and columns are obtained from weights.
the matrix P̃, expressed as vector D̃ and R̃, respectively, in Eq. (15). Wj com = ϕ · Wjsub + (1 − ϕ) · Wj
obj
(20)
Then, we can get the horizontal axis vector D̃ + R̃ and the vertical
axis vector D̃ − R̃. The IRM will be obtained by calculating the where ϕ is [0, 1] and determines the relative importance of
ordered pairs (D̃ + R̃, D̃ − R̃) [82]. objective and subjective weights.
⎧ ⎡ ⎤
n

⎪ ∑ 4.4. Later defuzzification VIKOR
⎨D̃ = (D̃i )n×1 = p̃ij ⎦

⎪ ⎣

j=1
[ n ] n×1 (15) VIKOR approach obtains the compromised solutions from the

⎪ ∑ ranking list, and the best alternative is determined based on
⎪R̃ = (R̃i )1×n = p̃ij priority ranking. Combined with the fuzzy set theory, some re-



i=1 1×n searchers proposed FVIKOR. Noting that although the defuzzifi-
cation step is critical in the FVIKOR method, the effect of de-
where D̃ and R̃ represent the sum of rows and columns of the fuzzification order on the decision-making results has not been
total-influence matrix, respectively. D̃i and R̃i represent the sum considered. In most previous studies, defuzzification is performed
of rows and columns of the total-influence matrix of ith criterion, before obtaining the best or worst performance value. However,
respectively. Chaghooshi indicated that the later the defuzzification step is
As mentioned previously, this study adopts the GMIR method completed, the more accurate the decision result will be [81].
to defuzzify the (D̃ + R̃, D̃ − R̃) to the crisp value ((D̃ + R̃)def , (D̃ − Under this consideration, in this study, we employ the fuzzy
R̃)def ). VIKOR approach in which the defuzzification step occurred at
a later stage, namely, LDVIKOR. The steps are summarized as
Step 5: Calculate the subjective weight of ith criterion Wisub .
follows:
[ ]1/2
((D̃i + R̃i )def )2 + ((D̃i − R̃i )def )2 Step 1: Collect the evaluation information using linguistic vari-
Wisub = ]1/2 (16) ables.
∑n [ This study uses the linguistic variables to represent the per-
i=1 ((D̃i + R̃i )def )2 + ((D̃i − R̃i )def )2
formance of each alternative, and each criterion describes the
6 H. Li, W. Wang, L. Fan et al. / Applied Soft Computing Journal 91 (2020) 106207

Fig. 1. The framework of the proposed fuzzy DEMATEL-EW-LDVIKOR approach.

judgments and preferences of the DMs. Table 5 lists the linguistic Table 5
Defined linguistic variables for evaluation ratings.
variables, which reflect the evaluation ratings of each option
Linguistic variables Abbreviations TFN
concerning all criteria [86].
Very poor VP (0, 0, 0.17)
Step 2: Construct the fuzzy decision matrix by aggregating the Poor P (0, 0.17, 0.33)
Medium poor MP (0.17, 0.33, 0.5)
performance evaluation of each DM based on Eq. (10). λk can be
Fair F (0.33, 0.5, 0.67)
adjusted according to the real scenarios. Medium good MG (0.5, 0.67, 0.83)
Good G (0.67, 0.83, 1)
Step 3: Calculate the fuzzy best value f˜j∗ = (xlij∗ , xm
ij , xij ) and fuzzy
∗ u∗
Very good VG (0.83, 1, 1)
worst value f˜j− = (xlij− , xm
ij , xij ) of jth criterion.
− u−

For Benefit criterion,


∗ For Benefit criterion,
f˜j = MAX (x̃ij ) = (max x̃lij , max x̃m
ij , max x̃ij )
u
j

j j j
(21) f˜j∗ − x̃ij
f˜j = MIN (x̃ij ) = (min x̃lij , min x̃m
ij , min x̃ij )
u d̃ij = (23)
j xuij∗ − xlij−
j j j

For Cost criterion,


For Cost criterion,
− x̃ij − f˜j∗
f˜j = MAX (x̃ij ) = (max x̃lij , max x̃m
ij , max x̃ij )
u
d̃ij = (24)
j
j j j xuij− − xlij∗
∗ (22)
f˜j = MIN (x̃ij ) = (min x̃lij , min x̃m
ij , min x̃ij )
u
j where d̃ij means the normalized fuzzy difference that Ci has on
j j j
Cj .
where the symbol * and – represent the best and worst value,
Step 5: Calculate the fuzzy group benefit value S̃ and the fuzzy
respectively. f˜j∗ and f˜j− mean the fuzzy best and worst value of
individual regret value R̃.
evaluation ratings that all criteria have on Cj . xlij∗ , xm ∗ u∗
ij , and xij de- n
note the best solution of the lower limit, middle limit, and upper ∑
S̃i = ⊕Wj com ⊗ d̃ij (25)
limit, respectively. xlij− , xm − u−
ij , and xij indicate the worst solution
j=1
of the lower limit, middle limit, and upper limit, respectively.
R̃i = MAX (Wj com ⊗ d̃ij ) (26)
Step 4: Compute the normalized fuzzy difference. j
H. Li, W. Wang, L. Fan et al. / Applied Soft Computing Journal 91 (2020) 106207 7

Table 6
Detailed expert information.
Expert Title Education Years of experience
E1 Senior Engineer Doctors in Mechanical Engineering 21
E2 Process Manager Bachelors in Process Engineering 12
E3 Operation Manager Masters in Industrial Engineering 15
E4 Process Engineer Masters in Mechanical Engineering 10
E5 Process Engineer Masters in Mechanical Engineering 5

Table 7 Table 8
Fuzzy direct-influence matrix of criteria given by five experts. Evaluation ratings of competitive alternatives.
Decision makers Criteria C1 C2 C3 C4 C5 C6 Decision-makers Alternative Criteria
C1 NI VL VL VH NI L C1 C2 C3 C4 C5 C6
C2 NI NI VH VH HI VL A1 VG G F P VG F
C3 NI VH NI VL NI NI A2 F MG F G MG VG
E1 E1
C4 VH NI VL NI L VL A3 VP P MP VG F VG
C5 VL VL L L NI VH A4 VG VG G F VG P
C6 VH L L VH VH NI
C1 NI L L VH VH VH A1 G MG MG P G MP
C2 VH NI VH VH VH VH A2 MG MG G MG MG G
E2
C3 VH VH NI VH NI L A3 P VP P VG MP VG
E2 A4 VG VG MG P G F
C4 VH VL VL NI VL HI
C5 VL VH L NI NI VH A1 VG F F P VG F
C6 VL NI L HI VL NI A2 F MG F MG MG G
E3
C1 NI VL VH VH HI HI A3 MP F MP VG F VG
C2 VH NI HI HI VH HI A4 VG G MG F G VP
C3 VL VH NI L HI VH
E3 A1 VG MG F P VG F
C4 VH HI HI NI HI L
A2 F MG F F MG G
C5 VH VL L VH NI HI E4
A3 MP P F VG F VG
C6 VL VL HI NI VL NI
A4 VG VG MG VP G F
C1 NI VH VH VH VH VH
C2 NI NI VH NI VH HI A1 VG F MG VP VG MP
C3 NI VH NI VL L HI A2 G MG F F MG G
E4 E5
C4 VH NI NI NI VL VL A3 VP F F VG P VG
C5 VL VH HI VL NI VH A4 VG G G VP VG F
C6 NI HI HI VL L NI
C1 NI HI HI VL L NI
C2 NI NI VH L HI VH
C3 VL VH NI HI L VL
When the following two conditions are satisfied, the alterna-
E5 tive can be ranked using Qi . The Qi of the optimal choice is the
C4 VH NI HI NI VL HI
C5 HI L L HI NI VH smallest.
C6 VL VL HI NI VL NI Cond1: Acceptable advantage
Q (A(2) ) − Q (A(1) ) ≥ 1/(N − 1) (28)

where S̃i = (Sil , Sim , Siu ) and R̃i = (Rli , Rm where Q (A(1) ) indicates the first alternative ranked by Qi , and
i , Ri ) represent the
u

fuzzy group benefit and individual regret value of ith alternative Q (A(2) ) represents the second alternative ranked by Qi .
(Ai ). Sil , Sim and Siu denote the group benefit of the lower limit, Cond2: Acceptable stability in evaluation
middle limit, and upper limit, respectively. Rli , Rm u
S(A(1) ) < {S(A(2) ), S(A(3) ), . . . , S(A(n) ) }
{ }
i and Ri mean
(29)
the individual regret value of lower limit, middle limit, and upper or R(A(1) ) < R(A(2) ), R(A(3) ), . . . , R(A(n) )
limit, respectively.
If both of the Cond1 and Cond2 are satisfied, the top alterna-
Step 6: Compute the fuzzy sorting index Q̃ . tive A(1) is the best.
If either of the above conditions is not satisfied, a set of
S̃i − S̃min R̃i − R̃min
Q̃i = v u l
⊕ (1 − v ) (27) compromise solution will be obtained.
Smax − Smax Rumax − Rlmax
• Alternative A(1) and A(2) if only Cond2 is not satisfied.
where Q̃i = (Qil , Qim , Qiu ) means the fuzzy sorting index of Ai . • Alternative A(1) , A(2) . . . , A(h) if only Cond1 is not satisfied,
Here, S̃min = MINi (S̃i ), Smaxu
= MAXi (Siu ), Smin
l
= MINi (Sil ), where h is the largest positive integer that satisfies Q (A(h) ) −
R̃min = MINi (R̃i ), Rumax = MAXi (Rui ) and Rlmin = MINi (Rli ). v is [0,1] Q (A(1) ) < 1/(N − 1).
and represents the weight of maximum group benefit, while 1-v
denotes the weight of individual regret. Generally, v is 0.5, which 4.5. The proposed hybrid fuzzy DEMATEL-EW-LDVIKOR
means the alternatives are evaluated in a consensus way [79].
Additionally, the smaller the value of Qi is, the higher the ranking To reduce the manufacturing cost and improve product qual-
of alternative represents. ity, a novel hybrid MCDM method, FDEMATEL-EW-LDVIKOR, is
employed to handle the machine tool selection process that sat-
Step 7: Defuzzify S̃i , R̃i and Q̃i using Eq. (11) into crisp value Si , Ri, isfies the multidimensional criteria. In particular, the proposed
and Qi . method utilizes the fuzzy set theory to represent the preferences
of the DMs, and fuzzy operators are utilized to obtain the fuzzy
Step 8: Rank the alternatives, sorting by values S, R, and Q in
initial direct-influence matrix. The FDEMATEL method is applied
ascending order.
to consider multicriteria interactions and compute the subjec-
Step 9: Obtain a compromise solution. tive weights of all criteria. Then, the hybrid method uses EW
8 H. Li, W. Wang, L. Fan et al. / Applied Soft Computing Journal 91 (2020) 106207

Fig. 2. The hierarchy for machine tool selection under multi-criteria.

technology to calculate the objective weights based on actual Table 9


performance data acquired in a real manufacturing process. Fur- Specific information about all alternatives.

thermore, this research develops the comprehensive weighting Criteria Type A1 A2 A3 A4

technique to integrate the subjective and objective weights. Fi- Maximum spindle speed (rpm) Benefit 12 000 24 000 3000 24 000
Failure rate (%) Cost 13% 14% 19% 9%
nally, LDVIKOR calculates the Q value, and the alternatives are
Utilization (%) Benefit 67% 71% 64% 74%
ranked according to the minimum Q value [87]. Fig. 1 shows the Max spindle torque (N m) Benefit 225 47 1200 87.5
framework of the proposed approach. Linkage accuracy (mm) Cost 0.34 0.18 0.41 0.21
Cost (RMB¥/h) Cost 250 600 200 700

5. Results and discussion

5.1. Data collection and scenarios setting matrix. Table 8 shows the evaluation ratings of competitive al-
ternatives based on the linguistic terms in Table 5. Meanwhile,
A practical case for machine tool selection is carried out in the real performance data for all criteria of the four options
this section. For confidentiality reasons, we cannot provide the are acquired by Distributed Numerical Control (DNC) in CITC for
name of the company for which the case study is performed. We, approximately one month, as presented in Table 9.
therefore, use the abbreviation ‘‘CITC’’ to denote the company,
which is located in Chengdu, China. This company focuses on 5.2. Comparison results
aircraft structural parts and automotive mold manufacturing. We
communicated with several senior process technicians in CITC 5.2.1. Subjective weight calculation
and discovered that it is challenging for them to select the proper The fuzzy initial direct-influence matrix is first constructed in
machine tool for a specific machining task under the constraints Table 10, where λE1 , λE2 , λE3 , λE4 , λE5 are 0.15, 0.15, 0.15, 0.25, 0.3,
of delivery time and processing quality. respectively. Then, we can calculate the fuzzy normalized direct-
influence matrix (Table 11). Table 12 shows the subjective weight
In general, the individuals who are particularly relevant for
W sub of the criteria.
the machine tool selection process should be involved in the
decision-making process. Therefore, the process engineers, ex-
5.2.2. Objective weight calculation
perts, and managers responsible for a machining task are asked
Table 13 shows the actual performance data L of each criterion
to evaluate the alternatives carefully. An ideal scheme of MCDM
for all alternatives derived from the DNC of CITC, and then the
problems needs a small group of three to five DMs [7]. This
normalized matrix V and entropy value Z are computed. Based
study invites two process engineers, two managers, and one on this, we can obtain the objective weight W obj of criteria in
senior engineer from CITC to provide advice. Thus, a committee Table 13.
comprising five DMs – designated E1, E2, E3, E4, and E5 – is
formed to realize the machine tool selection process. Table 6 lists 5.2.3. Later defuzzification VIKOR
the title, education, and years of experience of experts. After obtaining the subjective and objective weights, we can
Based on previous evaluation criteria explained in Section 2.1 derive the comprehensive weight (Table 14) under ϕ = 0.5.
and the actual situation in CITC, we selected the following deci- Afterward, the ranking results of all alternatives are acquired.
sion criteria, i.e., the maximum spindle speed, failure rate, uti- The evaluation rating information is presented. Then, Table 15
lization, minimum spindle torque, linkage accuracy, and cost, demonstrates the fuzzy best value and fuzzy worst value. Based
as shown in Fig. 2. For implementing one specific machining on this, S̃, R̃ and Q̃ are computed where v = 0.5 [82], which are
task, there are four candidates A1, A2, A3, and A4, as shown in defuzzified into the crisp values (Si , Ri , Qi ) using Eq. (11). Table 16
Fig. 2. Fig. 2 also shows the hierarchy structure, including the presents the rankings of alternatives by fuzzy group benefit value,
alternatives, criteria, and overall goal layers. individual regret value, and sorting index. Table 17 demonstrates
For each criterion, the linguistic variables determine the fuzzy the ranking results of S, R, and Q for all alternatives. From Eqs.
direct-influence of the five DMs, and Table 7 demonstrates its (30) and (31), since both Cond1 and Cond2 are satisfied, A4, which
H. Li, W. Wang, L. Fan et al. / Applied Soft Computing Journal 91 (2020) 106207 9

Table 10
Fuzzy initial direct-influence matrix of six criteria.
Criteria C1 C2 C3 C4 C5 C6
C1 (0,0,0.05) (0.08,0.13,0.16) (0.09,0.15,0.18) (0.15,0.2,0.2) (0.09,0.13,0.16) (0.11,0.16,0.19)
C2 (0.05,0.06,0.09) (0,0,0.05) (0.14,0.19,0.2) (0.08,0.11,0.15) (0.13,0.18,0.2) (0.11,0.16,0.19)
C3 (0.02,0.05,0.09) (0.15,0.2,0.2) (0,0,0.05) (0.06,0.11,0.15) (0.04,0.08,0.13) (0.06,0.09,0.14)
C4 (0.15,0.2,0.2) (0.02,0.03,0.08) (0.05,0.08,0.13) (0,0,0.05) (0.02,0.07,0.12) (0.05,0.10,0.15)
C5 (0.05,0.11,0.15) (0.08,0.13,0.16) (0.06,0.11,0.16) (0.06,0.10,0.15) (0,0,0.05) (0.14,0.19,0.2)
C6 (0.02,0.06,0.11) (0.03,0.08,0.13) (0.08,0.14,0.19) (0.04,0.07,0.11) (0.03,0.09,0.13) (0,0,0.05)

Table 11
Fuzzy normalized direct-influence matrix of six criteria.
Criteria C1 C2 C3 C4 C5 C6
C1 (0,0,0.05) (0.08,0.13,0.17) (0.11,0.16,0.19) (0.16,0.21,0.21) (0.10,0.14,0.17) (0.12,0.17,0.21)
C2 (0.05,0.06,0.11) (0,0,0.05) (0.15,0.20,0.21) (0.08,0.12,0.16) (0.14,0.19,0.21) (0.11,0.17,0.20)
C3 (0.02,0.06,0.10) (0.16,0.21,0.21) (0,0,0.05) (0.06,0.11,0.16) (0.05,0.08,0.13) (0.06,0.10,0.15)
C4 (0.16,0.21,0.21) (0.02,0.03,0.08) (0.05,0.09,0.14) (0,0,0.05) (0.02,0.08,0.13) (0.05,0.11,0.16)
C5 (0.06,0.11,0.15) (0.08,0.13,0.16) (0.07,0.12,0.17) (0.06,0.10,0.15) (0,0,0.05) (0.15,0.20,0.21)
C6 (0.02,0.06,0.11) (0.03,0.08,0.13) (0.09,0.14,0.20) (0.04,0.07,0.11) (0.04,0.09,0.13) (0,0,0.05)

Table 12
Subjective weight calculation result.
Criteria D̃ R̃ (D̃ + R̃)def (D̃ − R̃)def W sub
C1 (0.07,0.11,0.15) (0.55,1.03,1.36) 1.16 −0.89 0.1639
C2 (0.07,0.09,0.12) (−0.20,−0.66,−0.93) −0.54 0.72 0.1036
C3 (0.07,0.09,0.12) (0.73,0.71,0.65) 0.79 −0.61 0.1150
C4 (0.07,0.10,0.14) (1.18,1.72,1.98) 1.78 −1.57 0.2720
C5 (0.07,0.13,0.18) (−1.57,−1.91,−1.82) −1.71 1.97 0.2989
C6 (0.07,0.12,0.18) (−0.28,−0.23,−0.34) −0.13 0.38 0.0467

is the machine tool FIDIA GRT-28, is recommended as having Table 13


the superior performance to APEC GL-27, FIDIA Y2K-411, and Objective weight for each criterion.

FRFQ-250-VR/A8. Criteria C1 C2 C3 C4 C5 C6

Cond1: Acceptable advantage L 24 000 14 71 47 0.18 600


12 000 13 67 225 0.34 250
Q (A ) − Q (A ) = 0.443 − 0 ≥ 1/(4 − 1)
(2) (1)
(30) 3000 19 64 1200 0.41 200
24 000 9 74 87.5 0.21 700
Cond2: Acceptable stability V 1 0.5 0.7 0 1 0.2
{ 0.4286 0.6 0.3 0.1544 0.3043 0.9
S(A(1) ) < S(A(2) ), S(A(3) ), S(A(4) )
{ } 0 0 0 1 0 1
(31) 1 1 1 0.0351 0.8696 0
R(A(1) ) < R(A(2) ), R(A(3) ), R(A(4) )
{ }
Z 0.7479 0.7595 0.7203 0.3714 0.7206 0.6783
W obj 0.1259 0.1201 0.1397 0.3140 0.1396 0.1607
5.2.4. Comparison results with other existing methods
To verify the effectiveness of the proposed hybrid method, Table 14
we compare the findings of the proposed method with those of Comprehensive weight for each criterion.
existing methods proposed by EW-LDVIKOR (Approach 1) [88], Criteria C1 C2 C3 C4 C5 C6
FDEMATEL-LDVIKOR (Approach 2) [82], Fuzzy ANP-TOPSIS (Ap- W com 0.1449 0.1119 0.1274 0.2930 0.2173 0.1037
proach 3) [15]. When ϕ = 0, the calculation process complies
with Zhao’s method, called EW-LDVIKOR, wherein the objective
Table 15
weight determines the decision process. When ϕ = 1, the de- Fuzzy best and worst value of each criterion.
cision process is in correspondence with Chaghooshi’s approach, Criteria Fuzzy best value Fuzzy worst value
called FDEMATEL-LDVIKOR.
C1 (0.1660,0.2000,0.2000) (0.0136,0.0315,0.0652)
For Ayağ’s method, the best alternative can be determined C2 (0.1516,0.1847,0.2000) (0.0.297,0.0586,0.0918)
according to the preference rank order of closeness coefficient. C3 (0.1153,0.1484,0.1813) (0.0465,0.0799,0.1136)
The closeness coefficient of top priority alternative is the smallest. C4 (0.1660,0.2000,0.2000) (0.0000,0.0238,0.0564)
In this method, the parameter δ and µ are 0.5, and we use the five C5 (0.1612,0.1949,0.2000) (0.0414,0.0751,0.1085)
C6 (0.1660,0.2000,0.2000) (0.0462,0.0751,0.1088)
triangular fuzzy numbers (1̃, 3̃, 5̃, 7̃, 9̃) to describe the preference
in the pairwise comparison [15]. Then, we obtain the largest
eigenvalue λmax , consistency index (CI), consistency ratio (CR),
and relative weight of criteria Wrelativ e, as shown in Tables 18 5.3. Sensitivity analysis and discussion
and 19. Table 20 shows the evaluation criteria weigh Wcriteria by
synthesizing the normalized eigenvectors and relative weights of 5.3.1. Sensitivity analysis of parameter ϕ
six criteria. To evaluate the robustness of the proposed hybrid method,
Tables 21 and 22 show the sorting index of four approaches the sensitivity of the decision results under different values of
and the final ranking of four machine tool alternatives. Results of decision parameters ϕ is analyzed, and changes in the decision
four methods indicate that the top option is FIDIA GTF-28, and the results are observed. The scale unit of the parameter settings
worst choice is FRFQ-250-VR/A8, which verifies the effectiveness is 0.1, and this section considers a total of 11 scenarios. The ϕ
of the proposed method in machine tool selection. illustrates the weight preference of the DMs. If the DMs prefer
10 H. Li, W. Wang, L. Fan et al. / Applied Soft Computing Journal 91 (2020) 106207

Table 16
Results of fuzzy group benefit value, individual regret value and sorting index.
Alternative Fuzzy group benefit value Fuzzy group benefit value Fuzzy sorting index
A1 (0.4898, 0.4893, 0.4507) (0.3244, 0.3158, 0.3139) (0.4676, 0.4324, 0.4606)
A2 (0.5226, 0.5171, 0.4400) (0.3730, 0.3673, 0.3568) (0.5394, 0.5029, 0.4925)
A3 (0.8229, 0.8242, 0.7800) (0.6564, 0.6535, 0.6459) (1, 1, 1)
A4 (0.1487, 0.1651, 0.1199) (0.1145, 0.1292, 0.1028) (0, 0, 0)

Table 17 Table 19
Alternative ranking orders by S, R, and Q value. Consistency ratio calculation.
Alternative S R Q Prioritization Index λmax CI RI CR
S R Q Value 6.6262 0.1252 1.26 0.0994 < 0.1
A1 0.4830 0.3169 0.4430 2 2 2
A2 0.5052 0.3665 0.5073 3 3 3
Table 20
A3 0.8166 0.6527 1 4 4 4
Normalized eigenvectors and evaluation criteria weights.
A4 0.1548 0.1223 0 1 1 1
Criteria C1 C2 C3 C4 C5 C6 Wcriteria
C1 0.000 0.142 0.118 0.176 0.238 0.083 0.1889
Table 18 C2 0.250 0.000 0.059 0.294 0.334 0.000 0.1013
Pairwise comparison matrix for the relative weight of the criteria (δ = 0.5, C3 0.125 0.096 0.000 0.117 0.191 0.000 0.1099
µ = 0.5). C4 0.063 0.237 0.176 0.000 0.095 0.250 0.2394
Criteria C1 C2 C3 C4 C5 C6 Wrelativ e C5 0.188 0.191 0.235 0.354 0.000 0.667 0.2728
C6 0.374 0.334 0.412 0.059 0.142 0.000 0.0897
C1 1.000 1.000 1.000 1.000 3.000 5.000 0.1909
C2 0.750 1.000 9.000 3.000 5.000 7.000 0.4457
C3 0.750 0.113 1.000 5.000 3.000 3.000 0.1774
Table 21
C4 0.375 0.208 0.208 1.000 3.000 1.000 0.0761
Sorting index value of four approaches.
C5 0.375 0.208 0.375 0.375 1.000 3.000 0.0667
C6 0.208 0.146 0.375 0.750 0.375 1.000 0.0432 Alternative Q value Closeness coefficient
Proposed Approach 1 Approach 2 Approach 3
method
A1 0.443 0.377 0.508 0.678
the subjective weight, ϕ is set to 1. Conversely, if the DMs are A2 0.507 0.529 0.484 0.428
more concerned with the objective weight, where ϕ = 0. The Q A3 1 1 1 0.021
values in 11 scenarios are calculated when v = 0.5, as defined in A4 0 0 0.0243 0.979

Table 23. Fig. 3 shows the ranking orders of the four alternatives.
In general, the FIDIA GTF-28 is the most preferred alternative, and Table 22
FRFQ-250-VR/A8 is the least preferred. Alternative ranking of four approaches.
However, with the increase in ϕ , the Q value of A1 increased, Alternative Proposed method Approach 1 Approach 2 Appproach 3

while that of A2 decreased, thereby causing fluctuations in their A1 2 2 3 3


A2 3 3 2 2
rankings. As the differences of subjective weight and objective
A3 4 4 4 4
weight among C1, C2 and C3 are small, the main reason for the A4 1 1 1 1
change of ranking is the great difference of subjective weight and
objective weight among C4, C5 and C6. Especially, A1 is superior
to A2 in C4 and C6, and A2 is better than A1 in C5. The sum of
5.3.2. Discussion
weight difference of subjective and objective in C4 and C6 are
For this machining task, the actual ranking for these alterna-
0.156 and that in C5 is 0.159. Besides, as the contributions of tives is A4 > A1 > A2 > A3 provided by four senior decision
C4 and C6 for A1 is larger than that of C5 for A2, the change of makers of CITC (a real manufacturing company). According to
ranking for A1 and A4 occurs at ϕ = 0.8 not ϕ = 0.5. the decision-making results of four alternatives by the proposed
When ϕ is less than or equal to 0.8, the objective weight dom- MCDM model in this study, A4 (FIDIA GTF-28) is the best ma-
inates the decision result. Under this situation, A1 is better than chine tool to implement this task, and A3 (FRFQ-250-VR/A8) is
A2. The reason for this change is the advantage of A1 in maximum the worst choice. As alternative A1 has advantages in maximum
spindle torque (C4) and cost (C6). The objective weights of C4 and spindle torque and cost, and A2 has an advantage in linkage
C6 is high (0.3140 and 0.1604, respectively), while the subjective accuracy, the ranking of A1 and A2 will change when ϕ is greater
weight of this criterion is relatively low (0.2720 and 0.0467), and than 0.8.
A2 pays more attention to C5 that has a lower objective weight It can be seen from Table 22 that if the decision-making
(0.1396) than A1. Therefore, the Q of A1 is smaller than that of is only evaluated by expert knowledge, A2 will be the second
A2. choice, and A1 is inferior. In fact, for A2, due to the relatively
small maximum spindle torque, the product delivery time will
Similarly, when ϕ is less than or equal to 0.8, the decision
be delayed and the manufacturing cost will be high. Considering
result is dominated by objective weight. A1 is better than A2
the negative effects brought by A2, it will not be an alterna-
under this situation. The reason is probably that A2 has an ad-
tive option for managers in general, and the decision-making
vantage in max spindle torque (C5). The subjective weight of C5 result cannot achieve Pareto efficiency. On the contrary, if only
is high (0.2989), while the objective weight of this criterion is the objective data is utilized to evaluate the machine tool, the
relatively low (0.1396), and A2 pays more attention to C5 than decision-making results will overly rely on the actual operation
Aa. Therefore, as the relative importance of subjective weight data of machine tool and ignore the initiative of expert ex-
increases, the contribution of C5 is magnified, and C4 and C6 perience in decision-making, which makes the decision-making
are shrunken, which can explain the fluctuations in the ranking results difficult to adapt to the development trend of modern
orders of A2 and A1 in the sensitivity analysis. manufacturing industry.
H. Li, W. Wang, L. Fan et al. / Applied Soft Computing Journal 91 (2020) 106207 11

Table 23
Sorting index Q in 11 scenarios in terms of parameter ϕ .
ϕ 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
A1 0.377 0.391 0.403 0.416 0.429 0.443 0.457 0.469 0.482 0.495 0.508
A2 0.529 0.525 0.521 0.516 0.511 0.507 0.502 0.498 0.493 0.488 0.484
A3 1 1 1 1 1 1 1 1 1 1 1
A4 0 0 0 0 0 0 0 0 0 0 0

Fig. 3. Ranking orders of four alternatives under varied ϕ .

When considering both subjective and objective perspectives, more comprehensive. Secondly, this study only integrates DE-
A1 becomes the second alternative and is better than A2. This MATEL, EW, and VIKOR. Other weight methods (BWM, FUCOM)
result is consistent with the actual decision in CITC. The reason and ranking methods (TOPSIS) can be further explored to choose
is probably that the objective data can reflect the actual per- machine tools. Thirdly, the research utilizes the comprehensive
formance of the four alternatives. Therefore, the consideration weight technique to balance the subjective and objective weight.
of objective data will affect the comprehensive weight of each However, in the actual decision-making process, the historical
criterion, making the decision-making results closer to the actual performance data is critical in selecting the optimal machine tool.
situation. Thus, how to use more powerful weight approaches (like big data
Additionally, from the sensitivity analysis, we can notice that algorithms) to make the machine tool evaluation method more
no matter how the parameter ϕ that represent the preference of practical needs further study.
decision makers does change, alternative A4 always secures the Additionally, the presented study focuses on the manufactur-
first ranking and A3 is the last choice. It indicates that decision ing industry. The evaluation results may be different in other
results of machine tool selection by proposed fuzzy DEMATEL- fields, such as construction, service, transportation, etc., which
EW-LDVIKOR approach is effective and robust. This helps indus- need further research. Moreover, our research is static decision-
trial managers to choose the most appropriate machine tool to making, which limits our application in a dynamic environment.
implement machining tasks. With the development of artificial intelligence, it can help to
In summary, from the above analysis, the advantage of the design the intelligent decision-making system to support MCDM
evaluation model constructed in this study is that it can sci- problems, and the friendly human–computer interface will be
entifically and effectively integrate the expert knowledge with conducive to dynamic decision-making. Therefore, the integration
the actual operation performance of four alternatives, so that
of computer-based intelligence techniques into the ranking ap-
the decision results can not only reflect the expert’s experience
proaches should be further developed to make the MCDM system
but also coincide with the actual performance of the machine
more robust and intelligent.
tool, which is conducive to improving the overall performance of
the manufacturing system in CITC. Besides, the proposed fuzzy
6. Implication of the study
DEMATEL-EW-LDVIKOR model shows its advantage in flexibility
concerning the decision makers’ preference.
The major implications of the presented research have been
5.4. Limitations of the study described in two different sections. The first section discusses
the practical implications in the manufacturing field that has
Although the presented hybrid MCDM method has some ad- been brought from this research. The second section discusses the
vantages when seeking to select the best machine tool concerning social implications of the comprehensive benefits of the supplier
several criteria, there are still some limitations to be further chain of machine tools.
studied. Firstly, this research adopts only six evaluation criteria to
evaluate machine tools. However, with the increasing awareness 6.1. Managerial implication
of sustainability and green philosophy to manufacturing indus-
tries, the urgency of implementing green practice in the whole This study develops a novel hybrid MCDM model to select the
life cycle has become the focus of manufacturing and also be- best alternative. Firstly, the applications of linguistic variables and
come more imperative. Thus, we should explore more evaluation fuzzy set theory facilitate the collection of evaluation informa-
criteria to select machine tools, such as economic criteria, envi- tion from actual manufacturing factories and are conducive to
ronmental criteria, social criteria, making the evaluation method the mathematical operations of qualitative criteria. Secondly, the
12 H. Li, W. Wang, L. Fan et al. / Applied Soft Computing Journal 91 (2020) 106207

comprehensive weighting method integrating subjective based (3) Sensitivity analysis notices that no matter how parameter
on fuzzy DEMATEL with objective weight based on EW is uti- ϕ does fluctuate, the optimal alternative A4 and worst alternative
lized to deal with mixed criteria from subjective and objective A3 are not influenced. The analysis results indicate that the pro-
perspectives. Thirdly, our hybrid approach provides new insight posed method makes the decision making flexibly based on the
into MCDM problems, and it has an advantage of flexibility con- preference of the decision-makers.
cerning the decision-makers’ preference, which also offers a new The presented method can be applicable for supplier selec-
perspective for machine tool selection. tion and evaluation in the future, and the effective cooperation
Besides, this research contributes to the practice of machine with suppliers contributes to the sustainability achievement of
tool selection. The establishment of evaluation criteria and mea- the supply chain. We can evaluate the partners from economic,
surement of criterion weight provide a better understanding of environmental, social principles, and these principles are inter-
machine tool selection for manufacturing managers. Also, the dependent and conflicting. Then, we need to select the best
managers can select the best machine tool to complete their spe- alternative from all evaluated partners by ranking approaches.
cific machining tasks, which is conducive to improve machining Therefore, our hybrid model is a feasible method for supplier
efficiency, economic benefits, and product equality. Moreover, the selection.
weight of evaluation criteria provides some practical guidance While the presented hybrid MCDM approach provides some
for the design, service, sale, and maintenance of machine tools. meaningful guidance for selecting the best machine tool, there
Therefore, our method has substantial practical implications. are some limitations, including the incompletion of evaluation
criteria, the integration of other MCDM approaches, and the static
6.2. Social implication decision-making. Therefore, there are also further scopes that
can be included. Firstly, the more compressive evaluation criteria
With the rapid development of information technology such as (economic, environmental, and social criteria) and hybrid meth-
cloud computing, the machine tools of all enterprises are getting ods (BWM, FUCOM, and TOPSIS) can be investigated in machine
closer to the network. Cloud manufacturing aims to provide on- tool selection. Secondly, the friendly human–computer interface
demand manufacturing services to industrial managers according based on artificial intelligence will be developed to make the
to the service-oriented architecture. Through the cloud, machine MCDM model more intelligent and practical. Finally, the proposed
tools of manufacturing enterprises can be registered and shared. method can be extended into other fields such as service, supply
Especially for some small enterprises, this can help them get high- chain, and market.
valued machine tools to perform their machining tasks without
spending a lot of money to buy them. Our hybrid model provides Declaration of competing interest
some useful guidance for them to select the best machine tool
from other enterprises that meet their specific tasks. Thus, the No author associated with this paper has disclosed any po-
manufacturing industry chain becomes low energy consumption tential or pertinent conflicts which may be perceived to have
and high added value industries, and the shared and mutually impending conflict with this work. For full disclosure statements
beneficial manufacturing system will be formed gradually. There- refer to https://doi.org/10.1016/j.asoc.2020.106207.
fore, our hybrid approach has excellent social implications for
revitalizing manufacturing. CRediT authorship contribution statement
Additionally, the preferential tax and loan should be intro-
duced to provide financial and technological support for the Hai Li: Writing - original draft. Wei Wang: Writing - review &
construction and development of the shared platform of machine editing. Lei Fan: Validation. Qingzhao Li: Methodology. Xuezhen
tool selection, and the enterprises that explore and apply cloud Chen: Writing-review & editing.
manufacturing as well as generate corresponding products need
a large number of software developers and senior professional Acknowledgment
elites. Therefore, the presented model can alleviate the social
employment problem. The authors would like to thank the Major Project of National
Science and Technology (No. 2017ZX04002001), the Fundamental
7. Conclusion and future scope Research Funds for the Central Universities (ZYGX2019J032) and
NSAF (U1830110).
Engineers should carefully select the most suitable machine
tools to perform machining tasks, which is a complicated and References
time-consuming MCDM problem. Improper selection may lead
to loss in productivity and product quality, which may have [1] R. Sahal, J.G. Breslin, M.I. Ali, Big data and stream processing platforms for
a significant adverse impact on the overall performance of the industry 4.0 requirements mapping for a predictive maintenance use case,
manufacturing system. Therefore, a selection method of machine J. Manuf. Syst. 54 (2020) 138–151.
[2] Y. Xiong, J. Wu, C. Deng, Y. Wang, Machining process parameters opti-
tool is proposed. The main conclusions can be drawn as follows:
mization for heavy-duty CNC machine tools in sustainable manufacturing,
(1) The A4 alternative (FIDIA GTF-28) is optimal machine Int. J. Adv. Manuf. Technol. 87 (5–8) (2016) 1237–1246.
tool to implement this machining task, then closely followed by [3] L. Abdel-Malek, L.J. Resare, Algorithm based decision support system for
alternative A1 and A2. As the poor performance of most crite- the concerted selection of equipment in machining/assembly cells, Int. J.
ria, the A3 (FRFQ-250-VR/A8) alternative is last choice, which is Prod. Res. 38 (2) (2000) 323–339.
[4] R.B. Aronson, Remanufactured machine tools, Manuf. Eng. 131 (6) (2003)
more accurate than other methods when the objective data are
75–80.
considered. [5] Y. Du, L. Liao, L. Wang, Failure mode, effects and criticality analysis of
(2) The hybrid MCDM model, including fuzzy DEMATEL and remanufactured machine tools in service, Int. J. Precis. Eng. Manuf. 18 (3)
EW methods to determine criterion weights and LDVIKOR model (2007) 425–434.
to rank alternatives, is put forward. This model considers the [6] Z. Ayağ, R.G. Özdemir, An intelligent approach to machine tool selection
through fuzzy analytic network process, J. Intell. Manuf. 22 (2) (2011)
expert knowledge and actual performance of alternative, and en- 163–177.
ables decision makers to choose the most appropriate alternative [7] S. Önüt, S.S. Kara, T. Efendigil, A hybrid fuzzy MCDM approach to machine
under multiple conflicting criteria. tool selection, J. Intell. Manuf. 19 (4) (2008) 443–453.
H. Li, W. Wang, L. Fan et al. / Applied Soft Computing Journal 91 (2020) 106207 13

[8] J. Qin, X.W. Liu, W. Pedrycz, An extended VIKOR method based on prospect [36] R. Sadeghian, M.R. Sadeghian, A decision support system based on artificial
theory for multiple attribute decision making under interval type-2 fuzzy neural network and fuzzy analytic network process for selection of ma-
environment, Knowl-Based Syst. 86 (2015) 116–130. chine tools in a flexible manufacturing system, Int. J. Adv. Manuf. Technol.
[9] Z. Ayağ, A hybrid approach to machine-tool selection through AHP and 82 (9–12) (2016) 1795–1803.
simulation, Int. J. Prod. Res. 45 (9) (2007) 2029–2050. [37] A. Gabus, E. Fontela, World Problems an Invitation to Further Thought
[10] M. Liu, X. Zhang, H. Song, Z. Zhou, Inverse finite element method for re- Within the Framework of DEMATEL, Battelle Geneva Research Centre,
construction of deformation in the gantry structure of heavy-duty machine Switzerland, Geneva, 1972.
tool using FBG sensors, Sensors 18 (7) (2018) 2173. [38] J. Rezaei, Best-worst multi-criteria decision-making method, Omega 53
[11] K.C. Fan, H.M. Chen, T.H. Kuo, Prediction of machining accuracy (2015) 49–57.
degradation of machine tools, Precis. Eng. 36 (2) (2012) 288–298. [39] D. Pamučar, Ž. Stević, S. Sremac, A new model for determining weight
[12] M.C. Arslan, E. Budak, A decision support system for machine tool selection, coefficients of criteria in MCDM models: Full consistency method (FUCOM),
J. Manuf. Technol. Manage. 15 (1) (2004) 101–109. Symmetry 10 (9) (2018) 393.
[13] E. Çimren, B. Çatay, E. Budak, Development of a machine tool selection [40] M.J. Farrell, The measurement of productive efficiency, J. Roy. Statist. Soc.
system using AHP, J. Manuf. Technol. Manage. 35 (3–4) (2007) 363–376. Ser. A 120 (3) (1957) 253–281.
[41] B. Roy, Classement et choix en présence de points de vue multiples, Rev.
[14] A. Samvedi, V. Jain, F.T. Chan, An integrated approach for machine tool
Fr. Inform. Rech. Opér. 2 (8) (1968) 57–75.
selection using fuzzy analytical hierarchy process and grey relational
[42] T. Hill, R. Westbrook, SWOT analysis: it’s time for a product recall, Long
analysis, Int. J. Prod. Res. 50 (12) (2012) 3211–3221.
Range Plan. 30 (1) (1997) 46–52.
[15] Z. Ayag, R.G. Ozdemir, Evaluating machine tool alternatives through mod-
[43] D. Pamučar, L. Vasin, L. Lukovac, Selection of railway level crossings for
ified TOPSIS and alpha-cut based fuzzy ANP, Int. J. Prod. Econ. 140 (2)
investing in security equipment using hybrid DEMATEL-MARICA model, in:
(2012) 630–636.
XVI International Scientific-Expert Conference on Railway, Railcon, 2014,
[16] H.T. Nguyen, S.Z.M. Dawal, Y. Nukman, H. Aoyama, A hybrid approach
pp. 89–92.
for fuzzy multi-attribute decision making in machine tool selection with
[44] M. Keshavarz Ghorabaee, E.K. Zavadskas, Z. Z. Turskis, et al., A new
consideration of the interactions of attributes, Expert Syst. Appl. 41 (6)
combinative distance-based assessment (CODAS) method for multi-criteria
(2014) 3078–3090.
decision-making, Econ. Comput. Econ. Cybern. Stud. Res. 50 (3) (2016)
[17] M.H. Aghdaie, S.H. Zolfani, E.K. Zavadskas, Decision making in machine tool 25–44.
selection: An integrated approach with SWARA and COPRAS-G methods, [45] P.F. Tafreshi, M.H. Aghdaie, M. Behzadian, M.G. Abadi, Developing a group
Eng. Econom. 24 (1) (2013) 5–17. decision support system for advertising media evaluation: a case in the
[18] R. Rai, S. Kameshwaran, M.K. Tiwari, Machine-tool selection and operation middle east, Group Decis. Negot. 25 (5) (2016) 1021–1048.
allocation in FMS: solving a fuzzy goal-programming model using a genetic [46] B. Matić, S. Jovanović, D.K. Das, E.K. Zavadskas, Z. Stevic, S. Sremac, M.
algorithm, Int. J. Prod. Res. 40 (3) (2002) 641–665. Marinkovic, A new hybrid MCDM model: Sustainable supplier selection in
[19] S. Mishra, M.K. Prakash, R.S. Lashkari, A fuzzy goal-programming model of a construction company, Symmetry 11 (3) (2019) 353.
machine-tool selection and operation allocation problem in FMS: a quick [47] J. Szulecka, E.M. Zalazar, Forest plantations in Paraguay: Historical devel-
converging simulated annealing-based approach, Int. J. Prod. Res. 44 (1) opments and a critical diagnosis in a SWOT-AHP framework, Land. Use
(2006) 43–76. Policy 60 (2017) 384–394.
[20] M.H.M.A. Jahromi, R. Tavakkoli-Moghaddam, A novel 0-1 linear integer [48] S.H. Mousavi-Nasab, A. Sotoudeh-Anvari, A comprehensive MCDM-based
programming model for dynamic machine-tool selection and operation approach using TOPSIS, COPRAS and DEA as an auxiliary tool for material
allocation in a flexible manufacturing system, J. Manuf. Syst. 31 (2) (2012) selection problems, Mater. Des. 121 (2017) 237–253.
224–231. [49] İ. Kaya, M. Colak, F. Terzi, A comprehensive review of fuzzy multi criteria
[21] S. Perçin, H. Min, Optimal machine tools selection using quality function decision making methodologies for energy policy making, Energy Strateg.
deployment and fuzzy multiple objective decision making approach, J. Rev. 24 (2019) 207–228.
Intell. Fuzzy Syst. 24 (1) (2013) 163–174. [50] A. Imran, K. Nikolai, The assessment of regional economic potential based
[22] Y. He, Y. Li, T. Wu, J.W. Sutherland, An energy-responsive optimization on the methodology of fuzzy set theory, Procedia Comput. Sci. 120 (2017)
method for machine tool selection and operation sequence in flexible 372–375.
machining job shops, J. Clean. Prod. 87 (2015) 245–254. [51] Y.X. He, Y.X. Pang, Q. Zhang, Z. Jiao, Q. Chen, Comprehensive evaluation of
[23] P. Liu, J. Tuo, F. Liu, C. Li, X. Zhang, A novel method for energy efficiency regional clean energy development levels based on principal component
evaluation to support efficient machine tool selection, J. Clean. Prod. 191 analysis and rough set theory, Renew. Energy 122 (2018) 643–653.
(2018) 57–66. [52] R. Atta, M. Ghanbari, A high payload steganography mechanism based
[24] T.L. Saaty, What is the analytic hierarchy process, Math. Model. Decis. on wavelet packet transformation and neutrosophic set, J. Vis. Commun.
Support (1988) 109–121. Image Represent. 53 (2018) 42–54.
[25] S. Opricovic, Multicriteria Optimization of Civil Engineering Systems, [53] S. Mahmoudi, A. Jalali, M. Ahmadi, P. Abasi, N. Salari, Identifying critical
Faculty of Civil Engineering, Belgrade, 1998, pp. 5–21. success factors in Heart Failure Self-Care using fuzzy DEMATEL method,
[26] C.L. Hwang, K. Yoon, Multiple-Criteria Decision Making: Methods and Appl. Soft. Comput. 84 (2019) 105729.
Applications, A State of Art Survey, Springer-Verlag, New York, 1981. [54] W. Song, J. Cao, A rough DEMATEL-based approach for evaluating interac-
tion between requirements of product-service system, Comput. Ind. Eng.
[27] J.P. Vincke, P. Brans, A preference ranking organization method. The
110 (2017) 353–363.
PROMETHEE method for MCDM, Manage. Sci. (1985) 647–656.
[55] Z. Chen, X. Ming, X. Zhang, D. Yin, Z. Sun, A rough-fuzzy DEMATEL-ANP
[28] T.L. Saaty, Decision making with dependence and feedback: The analytic
method for evaluating sustainable value requirement of product service
network process, RWS Publ. (1996) 4922.
system, J. Clean. Prod. 228 (2019) 485–508.
[29] V. Keršuliene, E.K. Zavadskas, Z. Turskis, Selection of rational dispute
[56] Z. Chen, M. Lu, X. Ming, X. Zhang, T. Zhou, Explore and evaluate in-
resolution method by applying new step-wise weight assessment ratio
novative value propositions for smart product service system: A novel
analysis (SWARA), J. Bus. Econ. Manage. 11 (2) (2010) 243–258.
graphics-based rough-fuzzy DEMATEL method, J. Clean. Prod. 243 (2020)
[30] E.K. Zavadskas, A. Kaklauskas, Determination of an efficient contractor
118672.
by using the new method of multicriteria assessment, in: International [57] X.F. Ding, H.C. Liu, A 2-dimension uncertain linguistic DEMATEL method
Symposium for ‘‘The Organization and Management of Construction’’, in: for identifying critical success factors in emergency management, Appl.
Shaping Theory and Practice, vol. 2, 1996, pp. 94–104. Soft. Comput. 71 (2018) 386–395.
[31] M. Dağdeviren, Decision making in equipment selection: an integrated [58] M. Gul, E. Celik, N. Aydin, A.T. Gumus, A.F. Guneri, A state of the art
approach with AHP and PROMETHEE, J. Intell. Manuf. 19 (4) (2008) literature review of VIKOR and its fuzzy extensions on applications, Appl.
397–406. Soft. Comput. 46 (2016) 60–89.
[32] M.M. Yurdakul, Y.T. Iç, Analysis of the benefit generated by using [59] D. Liang, Y. Zhang, Z. Xu, A. Jamaldeen, Pythagorean fuzzy VIKOR ap-
fuzzy numbers in a TOPSIS model developed for machine tool selection proaches based on TODIM for evaluating internet banking website quality
problems, J. Mater. Process. Technol. 209 (1) (2009) 310–317. of ghanaian banking industry, Appl. Soft. Comput. 78 (2019) 583–594.
[33] A. Özgen, G. Tuzkaya, U.R. Tuzkaya, D. Özgen, A multi-criteria deci- [60] M. Abdel-Basset, A. Atef, F. Smarandache, A hybrid Neutrosophic multiple
sion making approach for machine tool selection problem in a fuzzy criteria group decision making approach for project selection, Cogn. Syst.
environment, Int. J. Comput. Int. Syst. 4 (4) (2011) 431–445. Res. 57 (2019) 216–227.
[34] Z. Taha, S. Rostam, A hybrid fuzzy AHP-PROMETHEE decision support [61] X. Zhang, J. Su, A combined fuzzy DEMATEL and TOPSIS approach for
system for machine tool selection in flexible manufacturing cell, J. Intell. estimating participants in knowledge-intensive crowdsourcing, Comput.
Manuf. 23 (6) (2012) 2137–2149. Ind. Eng. 137 (2019) 106085.
[35] M. Ilangkumaran, V. Sasirekha, L. Anojkumar, M. Boopathi-Raja, Ma- [62] R.K. Mavi, C. Standing, Critical success factors of sustainable project
chine tool selection using AHP and VIKOR methodologies under fuzzy management in construction: A fuzzy DEMATEL-ANP approach, J. Clean.
environment, Int. J. Model. Oper. Manage. 2 (4) (2012) 409–436. Prod. 194 (2018) 751–765.
14 H. Li, W. Wang, L. Fan et al. / Applied Soft Computing Journal 91 (2020) 106207

[63] D. Pamučar, M. Mihajlović, R. Obradović, P. Atanasković, Novel approach [75] Z. Taha, S. Rostam, A fuzzy AHP–ANN-based decision support system for
to group multi-criteria decision making based on interval rough numbers: machine tool selection in a flexible manufacturing cell, Int. J. Adv. Manuf.
Hybrid DEMATEL-ANP-MAIRCA model, Expert Syst. Appl. 88 (2017) 58–80. Technol. 57 (5–8) (2011) 719–733.
[64] S. Çali, S.Y. Balaman, A novel outranking based multi criteria group decision [76] J.H. Moon, C.S. Kang, Application of fuzzy decision making method to the
making methodology integrating ELECTRE and VIKOR under intuitionistic evaluation of spent fuel storage options, Prog. Nucl. Energy 39 (3–4) (2001)
fuzzy environment, Expert Syst. Appl. 119 (2019) 36–50. 345–351.
[65] L. Suganthi, Multi expert and multi criteria evaluation of sectoral invest- [77] G.S. Liang, M.J.J. Wang, A fuzzy multi-criteria decision-making approach
ments for sustainable development: An integrated fuzzy AHP, VIKOR/DEA for robot selection, Robot. Cim-Int. Manuf. 10 (4) (1993) 267–274.
methodology, Sustain. Cities. Soc. 43 (2018) 144–156. [78] S.K. Patil, R. Kant, A hybrid approach based on fuzzy DEMATEL and FMCDM
[66] Y. Wu, B. Zhang, C. Xu, L. Li, Site selection decision framework using fuzzy to predict success of knowledge management adoption in supply chain,
ANP-VIKOR for large commercial rooftop PV system based on sustainability Appl. Soft. Comput. 18 (2014) 126–135.
perspective, Sustain. Cities. Soc. 40 (2018) 454–470. [79] S. Opricovic, Fuzzy VIKOR with an application to water resources planning,
[67] M. Rajak, K. Shaw, Evaluation and selection of mobile health (mHealth) Expert Syst. Appl. 38 (10) (2011) 12983–12990.
applications using AHP and fuzzy topsis, Technol. Soc. 59 (2019) 101186. [80] H. Zhao, S. Guo, External benefit evaluation of renewable energy power in
[68] Y.A. Solangi, Q. Tan, N.H. Mirjat, S. Ali, Evaluating the strategies for China for sustainability, Sustainability. 7 (5) (2015) 4783–4805.
sustainable energy planning in Pakistan: An integrated SWOT-AHP and [81] A. Shemshadi, H. Shirazi, M. Toreihi, M.J. Tarokh, A fuzzy VIKOR method
Fuzzy-TOPSIS approach, J. Clean. Prod. 236 (2019) 117655. for supplier selection based on entropy measure for objective weighting,
[69] S. Boral, I. Howard, S.K. Chaturvedi, K. Mckee, V.N.A. Naikan, An integrated Expert Syst. Appl. 38 (10) (2011) 12160–12167.
approach for fuzzy failure modes and effects analysis using fuzzy AHP and [82] A.J. Chaghooshi, A. Arab, S.J.H. Dehshiri, A fuzzy hybrid approach for project
fuzzy MAIRCA, Eng. Fail. Anal. (2019) 104195. manager selection, Decis. Sci. Lett. 5 (3) (2016) 447–460.
[70] Z. Tian, J. Wang, H. Zhang, An integrated approach for failure mode and [83] C.E. Shannon, A mathematical theory of communication, Bell Syst. Tech. J.
effects analysis based on fuzzy best-worst, relative entropy, and VIKOR 27 (3) (1948) 379–423.
methods, Appl. Soft. Comput. 72 (2018) 636–646. [84] M.Z. Abidin, R. Rusli, A.M. Shariff, Technique for Order Performance by
[71] S.H. Zolfani, M.H. Aghdaie, A. Derakhti, E.K. Zavadskas, M.H.M. Varzabdeh, Similarity to Ideal Solution (TOPSIS)-entropy methodology for inherent
Decision making on business issues with foresight perspective; an appli- safety design decision making tool, Procedia Eng. 148 (2016) 1043–1050.
cation of new hybrid MCDM model in shopping mall locating, Expert Syst. [85] H. Guo, Research on term weighting algorithm based on information
Appl. 40 (17) (2013) 7111–7121. entropy theory, Comput. Eng. Appl. 49 (10) (2013) 140–146.
[72] M.H. Aghdaie, S.H. Zolfani, E.K. Zavadskas, Market segment evaluation and [86] C.C. Sun, A performance evaluation model by integrating fuzzy AHP and
selection based on application of fuzzy AHP and COPRAS-G methods, J. fuzzy TOPSIS methods, Expert Syst. Appl. 37 (12) (2010) 7745–7754.
Bus. Econ. Manage. 14 (1) (2013) 213–233. [87] S. Mandal, K. Singh, R.K. Behera, S.K. Sahu, N. Raj, J. Maiti, Human error
[73] A.I. Maghsoodi, H. Rasoulipanah, L.M. López, H.C. Liao, E.K. Zavadskas, identification and risk prioritization in overhead crane operations using
Integrating interval-valued multi-granular 2-tuple linguistic BWM-CODAS HTA, SHERPA and fuzzy VIKOR method, Expert Syst. Appl. 42 (20) (2015)
approach with target-based attributes: Site selection for a construction 7195–7206.
project, Comput. Ind. Eng. 139 (2020) 106147. [88] H. Zhao, H. Zhao, S. Guo, Evaluating the comprehensive benefit of eco-
[74] A.L. Zadeh, The concept of a linguistic variable and its application to industrial parks by employing multi-criteria decision making approach for
approximate reasoning—I, Inform. Sci. 8 (3) (1975) 199–249. circular economy, J. Clean. Prod. 142 (2017) 2262–2276.

You might also like