The Emerald Research Register for this journal is available at The current issue and full text archive

xt archive of this journal is available at

www.emeraldinsight.com/researchregister www.emeraldinsight.com/1355-2511.htm

JQME METHODOLOGY AND THEORY

11,4
Expert system for aircraft
maintenance services industry
348
Angus Cheung
China Aircraft Services Limited, The Hong Kong International Airport,
Lantau Island, Hong Kong, People’s Republic of China
W.H. Ip
Department of Industrial and Systems Engineering, The Hong Kong Polytechnic
University, Hong Kong, People’s Republic of China, and
Dawei Lu
Warwick Manufacturing Group, International Manufacturing Centre,
University of Warwick, Coventry, UK

Abstract
Purpose – The purpose of this study is to propose an approach to facilitate the allocation of labor
resources, which is a complex and fuzzy problem existing in the aircraft maintenance services
industry. On the other hand, the shortage of experienced and qualified engineers makes the labor
allocation process even more difficult.
Design/methodology/approach – Various approaches can be used to solve this
personnel-planning problem: mathematical programming is the common modeling approach;
however, it is found to be inappropriate where there are many intangible factors. The other approach is
analytical hierarchy process (AHP), a decision support method which can facilitate labor allocation. In
this paper, the authors propose a decision support system with fuzzy AHP in weighing the importance
of different intangible criteria; an extent analysis approach is used to overcome the uncertainty of
subjective perception.
Findings – In the allocation of labor resources, the personnel plans have to consider the aviation
authority regulations and safety laws, qualification of the employees’ and customers’ requirements, as
well as other intangible variables. The analysis results have shown that the fuzzy AHP approach-based
system provides better resource management and higher productivity for the aircraft industry.
Originality/value – Since limited studies have been found to be concentrated on manpower
allocation problems in the aircraft maintenance industry, this study can provide important findings
and references to the academicians and practitioners in the related areas.
Keywords Analytical hierarchy process, Aircraft, Maintenance, Resource management
Paper type Research paper

1. Introduction
In this paper, the authors describe an expert system that has been developing for
aircraft maintenance department of the China Aircraft Services Limited (CASL) at
Journal of Quality in Maintenance Hong Kong International Airport. CASL is one of the aircraft service providers in Hong
Engineering Kong. The services include ground supporting services, cabin services, aircraft
Vol. 11 No. 4, 2005
pp. 348-358 maintenance services and store and supply services. The aircraft maintenance services
q Emerald Group Publishing Limited include inspections for transit, turnaround and overnight aircraft, defects rectification,
1355-2511
DOI 10.1108/13552510510626972 B747 “A” and “B” checks, Cabin maintenance including in-flight entertainment
systems, engine change and outstation maintenance support. An important Expert system
management problem to be solved in this context is to guarantee that always for maintenance
sufficient engineers with appropriate qualifications are available at the airport for
carrying out the required services. The expert system that was developed for the services
maintenance department of CASL belongs to a class of decision support system that
focus on situations, where the workload of a service can be projected some time in
advance, based on a (preferably cyclic) time table and a set of norms specifying in 349
which time intervals and how much service has to be delivered by the department. The
proposed expert system can be used to select appropriate personnel and schedule the
personnel in order to meet the demand; the focus is therefore on the personnel selection
process.
Current engineer selection process is mainly relied on the decision-maker’s own
experience without systematic approach, which is apt to be biased due to limited
personal experience, knowledge and perception (Angus et al., 2002). The short lead
time very often results in the decision-maker making intuitive decisions; forget the
viable alternatives; cannot integrate the entire maintenance strategy and tactics of an
aircraft service company. It is not effective in obtaining an optimum decision. A better
approach is required to develop systematic and comprehensive labor-selection and
planning process, which can integrate relative importance among selection criteria and
precise identification of alternatives’ performance. The method of analytical hierarchy
process (AHP) is developed for a decision support system to assist the decision-making
of maintenance labor allocation: it has the inherent capacity to handle quantitative and
qualitative criteria used in labor selection (Tam and Rao, 2000). Further more it can
help improve the decision-making process by visualizing the problem systematically in
terms of criteria and sub-criteria (Tam and Rao, 2000). The decision-maker can
compare the alternatives’ performance against certain criteria using either pair-wise
comparison or a direct grade assignment. Nevertheless, AHP cannot deal with
uncertain problems precisely because it is usually hard to give discrete grades to
uncertain criteria, which lies within certain ranges with different degrees. In order to
overcome this problem, a fuzzy AHP with an extent analysis approach is proposed to
obtain the solution by assigning triangular fuzzy numbers to identify the relative
importance of criteria and alternatives’ weighting against some criteria.
The objectives of this paper is to describe and illustrate the use of fuzzy AHP
analysis approach to improve staff allocation as well as the support of decision-making
process within the maintenance industry. Through the fuzzy AHP analysis, a list of
labor and skill selected according to priority can be determined to perform a particular
maintenance task consistent with the real situation. In Section 2, we will review the
method and their application; the construction of the model is described in Section 3;
Sections 4 and 5 provide the discussion and conclusion.

2. Literature review
2.1 Analytic hierarchy process (AHP)
AHP proposed by Saaty (1980, 1994), has recently become increasingly popular in
dealing with multi-criteria decision problems, such as selecting machines for flexible
manufacturing systems (Tabucanon et al., 1994), evaluating the implementation of a
maintenance system (Labib et al., 1998) and vendor selection of a telecommunication
system (Tam and Rao, 2000).
JQME To apply AHP, a hierarchic model is constructed first. The simplest form consists of
11,4 three levels: the goal of the decision at the top level, followed by a second level
consisting of the criteria by which the alternatives, located in the third level, will be
evaluated step by step (Saaty and Vargas, 1994) (Figure 1). It can also be extended to a
more complex model by adding more sub-criteria under a certain level of criteria. The
model is given weightings of each alternative against the decision goal by evaluating
350 the importance of criteria and also weightings of each alternative against each
sub-criteria and criteria.
After constructing the hierarchic mode, the relative importance of each criteria
against the goal and weighting of each alternative against each criteria are determined
using pair-wise comparison using five-point scale of 1, 3, 5, 7, 9 as suggested by Saaty
and Vargas (1994) (Table I). A judgment matrix is formed for these evaluation criteria,
from which the eigenvectors are calculated and aggregated to measure the final
weighting of all decision alternatives. Finally, the alternatives are ranked according to
the weightings for decision-maker to make selection decision.

2.2 Fuzzy set theory

There are, however, some shortcomings connected with AHP approach suggested by Saaty.
Firstly, AHP is mainly used in nearly crisp decision applications (Chen, 1996; Hauser and
Tadikamalla, 1996). Secondly, because AHP only uses discrete scale of 1-9, it cannot take

Figure 1.
Simple hierarchic model of
AHP

Intensity of
importance Definition Explanation

1 Equal importance Two activities contribute equally to the objective

3 Weak importance of one Experience and judgment slightly favor one activity over
over another another
5 Essential or strong Experience and judgment strongly favor one activity over
importance another
7 Demonstrated importance An activity is strongly favored and its dominance
demonstrated in practice
9 Absolute importance The evidence favoring one activity over another is of the
Table I. highest possible order of affirmation
Intensity of importance
scale Source: Saaty and Vargas (1991)
into consideration the uncertainties connected with the decision-maker’s judgment (Cheng Expert system
and Mon, 1994). Moreover, the subjective judgment and preference of decision-maker’s have for maintenance
a strong effect in the AHP method (Cheng and Mon, 1994). To effectively overcome this
problem, the fuzzy logic principle is introduced in the AHP model. services
The fuzzy logic principle is based on a “superset” of Boolean logic that has been
extended to handle the concept of “partial truth” and it replaces the role of a
mathematical model using a number of rules with fuzzy variables and fuzzy terms 351
such as very hot, fairly cold, probably correct (Buchanan and Shortliffe, 1984; Leung
and Lam, 1988; Orchard, 1994). This usually occurs when there is neither definite
quantitative description nor boundaries to a certain object. For example, terms such as
hot, old, short are fuzzy: a person of 60 can be described to be “old” compared to a
person of 30. However, it does depend on the context where it is considered. Unlike
classical set theory, which handles with clearly defined membership to a set, fuzzy
membership of an element to a set can be partial with the element belonging to a set
based on a certain grade of membership (normally from 0 to 1) (Lee et al., 2001).
In mathematical terms, fuzzy set A is defined in a relevant universal set X by a
membership function, which assigns to each element x of X a number, A (x), in the
closed unit interval [0, 1] that characterized the degrees of membership of x in A.
Membership functions are thus functions of the form (Klir et al., 1997):
A:X ½0; 1

Supported by fuzzy set theory, triangular-shaped membership function, characterized
by three parameters, l, m, and n, as shown in Figure 2 are used to assign weightings to
alternatives’ performance against criteria, which could represent the uncertainty and
gradual level. Besides, they are defined in equation (1) for triangular fuzzy numbers x~ :
8
>
> 1 x¼m
>
>
>
< x 2 l 1 ,¼ x ,¼ m
mðxÞ ¼ m 2 l ð1Þ
>
> n 2 x
>
>
: n 2 m m ,¼ x ,¼ n
>

2.3 Fuzzy AHP method

In fuzzy AHP the hierarchic model is constructed in the same way as AHP. But the
scoring method is different; the ranking method is also different. When the criteria are
compared to each other using the five-point scale, a fuzzy triangular number instead of

Figure 2.
Triangular membership
functions
JQME crisp number is used to give score to describe the fuzzy importance level. For example,
11,4 “approximately equally important” can be expressed with a fuzzy set, which also
includes from 1 (equal importance) to 3 (weak importance) with different levels of
memberships. The detailed membership functions of fuzzy number are listed in
Table II.
The alternatives are also assigned triangular fuzzy number of 1; ~ 3; ~ 5;
~ 7; ~ 9~ to
352 measure their performance against to each criteria. Either pair-wise comparison or
direct assignation is used. At the same time, fuzzy ratio scales for criteria should be
defined to transfer the quantitative performance to corresponding fuzzy number.
If we assume that the fuzzy weight vector W of the selection criteria is ½W ~ j
1£n and
the fuzzy judgment matrix A of alternatives ½A1 ; A2 ; . . .Am
is ½~aij
m£n ; then the final
score P of alternatives can be calculated as follows:

P ¼ A^W T ð2Þ

0 1 0 1
a~ 11 a~ 12 ... a~ 1n w~ 1
B C B C
B a~ 21 a~ 22 ... a~ 2n C B w~ 2 C
B C B C
¼B C^B C
B ... C B...C
@ A @ A
a~ m1 a~ m2 . . . a~ mn w~ n

0 1
a~ 11 ^w~ 1 %~a12 ^w~ 2 %. . .%~a1n ^w~ n
B C
B ... C
B C
¼B C
B ... C
@ A
a~ m1 ^w~ 1 %~am2 ^w~ 2 %. . .~amn ^w~ n

0 1
r~1
B C
B r~2 C
B C
¼B C
B...C
@ A
r~m

Fuzzy number Definition Membership function

1~ Equal importance (1, 1, 3)

Table II. 3~ Weak importance (1, 3, 5)
The definition and 5~ Strong importance (3, 5, 7)
membership function of 7~ Demonstrated importance over the other (5, 7, 9)
fuzzy number table 9~ Absolute importance (7, 9, 9)
Addition and multiplication for the fuzzy numbers in the above equations are stated in Expert system
the following:
Fuzzy number addition %
for maintenance
services
~ B~ ¼ ½a1 þ b1; a2 þ b2; a3 þ b3

A% ð3Þ

Fuzzy number multiplication ^ 353

~ B~ ¼ ½a1 £ b1; a2 £ b2; a3 £ b3

A^ ð4Þ

In order to rank the final fuzzy scores of alternatives, a crisp total ordering from fuzzy
numbers are constructed. In this study, we selected the fuzzy mean and spread method
to defuzzify and rank the fuzzy number since human intuition would favor a fuzzy
number with higher mean value and at the same time lower spread (Lee et al., 2001):

1
Mean x~ ð~ri Þ ¼ ðl þ m þ nÞ ð5Þ
3

1 2
Standard deviation s~ð~ri Þ ¼ ðl þ m 2 þ n 2 2 lm 2 ln 2 mnÞ ð6Þ
18

3. The proposed model

The conventional approaches in personnel selecting process of aircraft maintenance
service can be determined according to the managers’ and duty managers’ judgment
based on their knowledge and experiences and checklist method (Angus et al., 2002).
These approaches can only provide a set of systematic steps for problem solving
without involving the relationship among the decision factors. Meanwhile, the ability
and experience of the analyst(s) may also significantly influence the performance of the
final result. Therefore, we have enhanced the decision process using fuzzy sets theory
to integrate with AHP model. The proposed system consists of four stages:
(1) construct hierarchical structure for fuzzy AHP;
(2) weights determination;
(3) data collection; and
(4) decision making.

They are elaborated in the following section.

3.1 Construct hierarchical structure for fuzzy AHP

The goal is to select an optimal staff for a particular maintenance task. In order to
achieve this goal, several criteria are used to in the selection process, which are mainly
based according to license qualifications from, airline company, aviation authority and
company as well as personal experience. Because the criteria of license qualifications
and personal experience are the most critical, they are included in the model first. The
following diagram illustrates the selection criteria (Figure 3).
JQME
11,4

354

Figure 3.
Selection criteria

3.2 Weight determination

According to the prevailing situation in the maintenance industry and a certain airline
service company, the criteria’s weights are directly assigned using crisp number of 1, 3,
5, 7, 9 instead of pair-wise comparison because it would simplify the problem as well as
to provide sufficient accuracy. Tables III and IV show the detailed weighting of each
criterion.

Criteria Description Weighting

AME license Aircraft maintenance engineer license 9

Airplane approval Approval to perform maintenance for a particular airplane
type, e.g. Boeing series airplanes 9
Airlines approval Approval from a particular airline company 9
Years Years of experience doing maintenance tasks 5
Specialization Which parts of maintenance specialization, e.g. airframe,
engine, electrical, radio, etc. 3
Airplane experience Years of experience doing maintenance for a particular
Table III.
airplane type 5
Description and
weighting of criteria in Airlines experience Years of experience doing maintenance for a particular
the first selection process airline company 5

Criteria Description Weighting

Table IV.
Description and Training Whether the alternative has had sufficient training 5
weighting of criteria in Regulations Whether the alternative follows company regulations 7
the second selection Shift time Whether the alternative is available in that time slot 3
process Human factor Whether the alternative has done too long work hours 1
3.3 Data collection Expert system
In this case study, the alternatives of size 40 are given, respectively, to the weightings for maintenance
against each criterion. Again, they are assigned straightforward because there are too
many alternatives (40) to make comparison manually through a 40-40 matrix. services
For criteria – AME license, airplane approval and airlines approval, alternatives are
assigned either crisp 9 or 1 because they can only be either a license holder or not.
On the other hand, fuzzy number is used against other criteria – years, 355
specialization, airplane experience, and airlines experience. The fuzzy ratio scales for
criteria of years, airplane experience and airlines experience are stated in Table V.

4. Evaluations and discussion

During implementation in the case study, the final score P of alternatives is calculated
using equation (2). It is defuzzfied according to equations (5) and (6), the mean and the
standard deviation of P are also calculated. Because the mean is found not to be equal
to each other, they are used to rank alternatives according to equation (5). The detail
calculation is shown in Figure 4.
From the results it is found that the first six alternatives have highest scores, they
are selected for the second selection process, which involve the company regulation
criteria – training, regulations, shift time and human factor. Using the same method in
the first selection process, the alternatives are directly assigned scores in the form of
fuzzy triangular number against these criteria. Then the multiplication results are
defuzzified; their mean are ranked in Figure 5.
From the figure it can be seen that alternative 6 has the highest score. In the case
study, we inputted a list of flight schedule into the system. The system will match the
schedule maintenance task automatically from database and add new task into the
scheduled flight. This expert system will perform the analysis and generate a
personnel plan to meet the maintenance task. It ensures that the personnel plan has
appropriate staff and qualifications to perform the maintenance task according to
customer requirements.
In the evaluation of the model we have performed sensitivities analysis, this is
conducted by changing the input criteria weighting and input of alternatives’
weighting; both cases will change the ranking of the result. The result shows that the
top three alternatives change slightly with each criteria weighting. The overall
weighting against all criteria are relatively stable and rarely affected by weighting
variation of individual criteria. The expert system using fuzzy AHP can represent
uncertain weightings of criteria and alternatives in a gradual manner. It is robust and
accurate enough for us to determine appropriate personnel for the maintenance task. It
is flexible and effective to cater for many more alternatives and criteria in the selection
process. Moreover, further research is being carried out to integrate the personnel plan
produced by the expert system for maintenance jobs despatching and completion in the
operation, equipment and material scheduling.

5. Conclusion
In the industry of aircraft maintenance, the maintenance personnel allocation is a
complicate and important issue. How to select a suitable staff to perform a particular
maintenance task at the right time is a critical factor for the success of maintenance
service company. In this paper, a fuzzy AHP approach is proposed to facilitate this
JQME
11,4

356

Figure 4.
Ranking of 40 alternatives
after first selection process
 Expert system
for maintenance
services

357
Figure 5.
Ranking of six alternatives
after second selection
process

Scales Years Airplane experience Airlines experience

1~ 1-3 1-2 1-2

3~ 4-6 3-4 3-4
5~ 7-9 5-6 5-6
7~ 10-12 7-8 7-8 Table V.
9~ 13-15 9-10 9-10 Fuzzy ratio scales

selection process as well as improve its reliability and accuracy. The case study is used
to illustrate the selection of most suitable engineer for a particular maintenance task
using a fuzzy judgment matrix. This expert system can be extended to support and
integrate with other equipment, material and operation processes. The results indicate
that the system provides better resource management and higher productivity for the
aircraft industry.

References
Angus, H.W.C., Dawei, L.U. and Ip, W.H. (2002), Proceedings of 3rd International Conference on
Quality and Reliability: Defining Quality and Reliability in the Digital Era, Melbourne.
Buchanan, B.G. and Shortliffe, E.H. (1984), Rule-based Expert Systems, Addison-Wesley,
Reading, MA.
Chen, S.M. (1996), “Evaluating weapon systems using fuzzy arithmetic operations”, Fuzzy Sets
and Systems, Vol. 77, pp. 265-76.
Cheng, C.H. and Mon, D.L. (1994), “Evaluating weapon system by AHP based on fuzzy scale”,
Fuzzy Sets and Systems, Vol. 63, pp. 1-10.
Hauser, D. and Tadikamalla, P. (1996), “The analytic hierarchy process in an uncertain
environment: a simulation approach”, European Journal of Operational Research, Vol. 91
No. 1, pp. 27-37.
Klir, G.J., St Clair, U.H. and Bo, Y. (1997), Fuzzy Set Theory Foundations and Applications,
Prentice-Hall PTR, Inc., Englewood Cliffs, NJ.
Labib, A.W., O’Connor, R.F. and Williams, G.B. (1998), “Effective maintenance system using the
analytic hierarchy process”, Integrated Manufacturing Systems, Vol. 9 No. 2, pp. 87-98.
JQME Lee, W.B., Lau, H., Liu, Z.Z. and Tam, S. (2001), “A fuzzy analytic hierarchy process approach in
modular product design”, Expert Systems, Vol. 18 No. 1, pp. 32-42.
11,4 Leung, K.S. and Lam, W. (1988), “Fuzzy concepts in expert systems”, Computer, Vol. 21 No. 9,
pp. 43-56.
Orchard, A. (1994), FuzzyCLIPS Version 6.02A User’s Guide, National Research Council of
Canada, Ottawa.
358 Saaty, T.L. (1980), The Analytic Hierarchy Process, McGraw-Hill, New York, NY.
Tabucanon, M.T., Batanov, D.N. and Verma, D.K. (1994), “Decision support system for
multi-criteria machine selection for flexible manufacturing systems”, Computers in
Industry, Vol. 25 No. 2, pp. 131-43.
Tam, C.Y. and Rao, V.M. (2000), “An application of the AHP in vendor selection of a
telecommunications system”, Omega: The International Journal of Management Science,
Vol. 29 No. 2001, pp. 171-82.

Further reading
Alfares, H.K. (1999), “Aircraft maintenance workforce scheduling: a case study”, Journal of
Quality in Maintenance Engineering, Vol. 5 No. 2, pp. 78-88.
Triantaphyllou, E. (2000), Multi-criteria Decision-Making Methods: A Comparative Study, Kluwer
Academic Publishers, Dordrecht.