Hindawi

Complexity
Volume 2019, Article ID 1046054, 8 pages
https://doi.org/10.1155/2019/1046054

Research Article
Fault Tree Interval Analysis of Complex Systems Based on
Universal Grey Operation

Feng Zhang ,1 Shiwang Tan,1 Leilei Zhang,2 Yameng Wang,1 and Yang Gao1
1
School of Mechanics, Civil Engineering and Architecture, Northwestern Polytechnical University, Xi’an 710129, China
2
No. 203 Research Institute of China Ordnance Industries, Xi’an 710065, China

Correspondence should be addressed to Feng Zhang; [email protected]

Received 24 July 2018; Revised 1 November 2018; Accepted 4 December 2018; Published 1 January 2019

Academic Editor: Xinggang Yan

Copyright © 2019 Feng Zhang et al. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

The objective of this study is to propose a new operation method based on the universal grey number to overcome the shortcomings
of typical interval operation in solving system fault trees. First, the failure probability ranges of the bottom events are described
according to the conversion rules between the interval number and universal grey number. A more accurate system reliability
calculation is then obtained based on the logical relationship between the AND gates and OR gates of a fault tree and universal grey
number arithmetic. Then, considering an aircraft landing gear retraction system as an example, the failure probability range of the
top event is obtained through universal grey operation. Next, the reliability of the aircraft landing gear retraction system is evaluated
despite insufficient statistical information describing failures. The example demonstrates that the proposed method provides many
advantages in resolving the system reliability problem despite poor information, yielding benefits for the function of the interval
operation, and overcoming the drawback of solution interval enlargement under different orders of interval operation.

1. Introduction The traditional fault tree analysis method is based on a

probability model; when there is a large set of failure samples
The fault tree analysis (FTA) method is typically applied as and other sufficient statistical information describing the
the main method in the reliability analysis of large systems evaluated parts of a system, the uncertainty of bottom events
[1–3]. A fault tree is a logical block diagram composed of a can be quantified independently [5, 12]. However, when only
top event (outcome), intermediate events, and bottom events small sample sets are available in an engineering analysis case,
and is used to describe the internal functional logical rela- the statistics describing component failure are insufficient
tionship between events. The logical relationships between to accurately estimate failure distribution [13, 14]. It is thus
the components of a system and their events are obtained difficult to determine the failure probability of components
based on the operating principle and fault mechanisms of in many kinds of complex systems, such as landing gear
the system. The top-level failure probability of the system retraction systems and large-scale space-borne antennae.
can be obtained through the logical relationship between This limits the application of the probabilistic model-based
event layers and through data operations using the failure fault tree method in complex engineering applications.
probability statistics of the underlying components of the Based on the fuzzy set theory, the fuzzy fault tree
system. Then, a system-level reliability evaluation can be describes the probability of event occurrence using various
performed [4–7]. In recent studies, researchers have made fuzzy numbers, addressing the difficulties associated with
major achievements in theoretical system and engineering precisely measuring the probability of base event occurrence
reliability analyses based on fault trees. The FTA method due to the complexity of the environment and incomplete
based on a probability model has seen wide application in data [15–18]. Ding and Lisianski regarded the performance
several systems engineering fields such as aviation, aerospace, rate and corresponding probability of an event as fuzzy
and nuclear power [8–12]. values and developed a reliability evaluation technique for
2 Complexity

a multistate system using the fuzzy universal generation [𝑦] = [𝑦, 𝑦] ∈ 𝐼(𝑅), the binary operation on interval set 𝐼(𝑅)
function [19]. Li et al. introduced random fuzzy variables is defined as follows:
and proposed a hybrid universal generation function [20].
Liu and Huang proposed a fuzzy continuous-time Markov [𝑥] ∗ [𝑦] = {𝑧 | 𝑧 = 𝑥 ∗ 𝑦, 𝑥 ∈ [𝑥] , 𝑦 ∈ [𝑦]} (1)
model with a finite discrete state and used it to evaluate
the fuzzy state probability of multistate elements at any time The four arithmetic operation rules can then be derived
[21]. However, in the fuzzy fault tree, the determination of as follows [27]:
the fuzzy value, fuzzy variable, and fuzzy state probability is
highly subjective. [𝑥] + [𝑦] = [𝑥 + 𝑦, 𝑥 + 𝑦]
The interval domain is an important model in non-
probability theory: the shape of an interval domain rep- [𝑥] − [𝑦] = [𝑥 − 𝑦, 𝑥 − 𝑦]
resents the degree to which events occur in an interval
model, while the size of the interval domain signifies the [𝑥] ⋅ [𝑦]
volatility or degree of deviation of an uncertain event. To (2)
establish an interval model, only the boundaries of an event = [min (𝑥𝑦, 𝑥𝑦, 𝑥𝑦, 𝑥 𝑦) , max (𝑥𝑦, 𝑥𝑦, 𝑥𝑦, 𝑥 𝑦)]
set are required, not its internal distribution. This results
in significant independence from the data compared to a [𝑥] 1 1
= [𝑥, 𝑥] ⋅ [ , ] (0 ∉ [y])
conventional probability model [22–27]. However, it should [𝑦] 𝑦 𝑦
be pointed out that the power exponentiation of an interval
number will lead to the expansion of the interval and It can be seen in (2) that the calculation of the interval
that different orders of operation performed on the same number provides an extremely wide range due to the influ-
interval numbers can provide different expansion intervals ence of interval expansion. This is the primary drawback of
[28, 29]. the interval method.
The universal grey number provides the function of the
interval operation and overcomes the drawback associated 2.2. Four Arithmetic Operations of Interval Analysis Based on
with traditional interval operation, i.e., the change in solution the Universal Grey Operator. Setting the domain as 𝑈 = 𝑅
interval with order of operation [29]. Some scholars have (the set of real numbers), the universal grey number set in 𝑅
gradually introduced and successfully applied the universal is denoted by 𝑔(𝑅). Calling an element in 𝑔(𝑅) the universal
grey operation to structural reliability research [30–33]. Luo grey number, 𝑔 = (𝑥, [𝜇, 𝜇]), 𝑥 ∈ 𝑅, and 𝜇, 𝜇 ∈ 𝑅, where 𝑥 is
introduced the grey range transformation into the process
of model building to eliminate the incomparability of dif- the observed value and [𝜇, 𝜇] is the grey information portion
ferent dimensions and achieved an effective risk assessment of 𝑥.
of the ice plug phenomenon [30]. Jin et al. proposed a The corresponding four arithmetic operation rules are
generalized Rayleigh quotient method based on general- accordingly [27]
ized grey mathematics to represent the interval parameters
𝑥1 𝜇1 + 𝑥2 𝜇2 𝑥1 𝜇1 + 𝑥2 𝜇2
in uncertain structures using generalized grey numbers 𝑔1 + 𝑔2 = (𝑥1 + 𝑥2 , [ , ]) (3)
[31]. Liu et al. considered the uncertainty of the inter- 𝑥1 + 𝑥2 𝑥1 + 𝑥2
val arithmetic for the structural, nonprobabilistic reliability 𝑥1 𝜇1 − 𝑥2 𝜇2 𝑥1 𝜇1 − 𝑥2 𝜇2
calculation of nonlinear systems, using the universal grey 𝑔1 − 𝑔2 = (𝑥1 − 𝑥2 , [ , ]) (4)
number instead of the interval parameters to overcome the 𝑥1 − 𝑥2 𝑥1 − 𝑥2
impact of interval arithmetic uncertainty on reliability results
[32]. 𝑔1 × 𝑔2 = (𝑥1 𝑥2 , [𝜇1 𝜇2 , 𝜇1 𝜇2 ]) (5)
Based on the advantages of the universal grey number 𝜇1 𝜇
𝑔1 𝑥
method, a new method for solving the reliability of the top = ( 1 , [ , 1 ]) (6)
event of a fault tree is proposed in this paper to overcome 𝑔2 𝑥2 𝜇2 𝜇2
the shortcomings of the existing nonprobabilistic reliability
method of interval operation. The proposed method complies In practical applications, the universal grey number and
with the conversion rule between the interval number and interval number can be interchanged with each other via
universal number, and the four arithmetic operations of the conversion. For the grey number 𝑔 = (𝑥, [𝜇, 𝜇]), the
universal grey number. corresponding interval number is in the form of 𝑔 = [𝑥𝜇, 𝑥𝜇].
The interval number 𝑔 = [𝑥, 𝑥] can be uniformly expressed
as 𝑔 = (𝑥, [𝑥/𝑥, 1]) for ease of operation.
2. Interval Analysis of Simple System
Fault Trees 2.3. Interval Analysis of Fault Trees Based on Universal Grey
Operator for an OR Gate Operator. A fault tree interval
2.1. Four Arithmetic Operations of Traditional Interval Anal- analysis based on the universal grey operator is performed
ysis. Let ‘∗’ represent a real binary operation on the set of by using an OR gate operator with three bottom events as an
real numbers, ∗ ∈ {+, −, ⋅, /}. For [𝑥] = [𝑥, 𝑥] ∈ 𝐼(𝑅) and example. The fault tree is shown in Figure 1.
Complexity 3

T The failure probability intervals of the three bottom

events can then be expressed in terms of universal grey
+
numbers as
1
𝑃1 = (0.3, [ , 1]) (18)
3
P1 P2 P3
1
𝑃2 = (0.6, [ , 1]) (19)
Figure 1: OR gate operator with three bottom events. 3
1
𝑃3 = (0.8, [ , 1]) (20)
2
The failure function of the top event in relation to the The interval operation based on the universal grey oper-
bottom event is given by ator is performed by substituting (18) to (20) into (7) to (10),
and the probability interval ranges of the top event of the fault
𝑃𝑇1 = 1 − (1 − 𝑃1 ) (1 − 𝑃2 ) (1 − 𝑃3 ) (7) tree in the four forms are as follows:
1
𝑃𝑇1 = (1, [1, 1]) − {(1, [1, 1]) − (0.3, [ , 1])}
and the other three equivalent forms are given as follows: 3
1
⋅ {(1, [1, 1]) − (0.6, [ , 1])} ⋅ {(1, [1, 1])
𝑃𝑇2 = 1 − (1 − 𝑃1 − 𝑃2 + 𝑃1 𝑃2 ) (1 − 𝑃3 ) (8) 3 (21)
𝑃𝑇3 = 1 − (1 − 𝑃1 ) (1 − 𝑃2 − 𝑃3 + 𝑃2 𝑃3 ) 1 54
(9) − (0.8, [ , 1])} = (1, [1, 1]) − (0.056, [ , 1])
2 7
𝑃𝑇4 = 𝑃1 + 𝑃2 + 𝑃3 − 𝑃1 𝑃2 − 𝑃1 𝑃3 − 𝑃2 𝑃3 + 𝑃1 𝑃2 𝑃3 (10)
= [0.568, 0.944]

The failure probabilities of the three bottom events are 1

𝑃𝑇2 = (1, [1, 1]) − {(1, [1, 1]) − (0.3, [ , 1])
expressed by the interval number as follows: 3
1 1 1
− (0.6, [ , 1]) + (0.3, [ , 1]) ⋅ (0.6, [ , 1])}
𝑃1 = [0.1, 0.3] (11) 3 3 3
(22)
𝑃2 = [0.2, 0.6] (12) 1
⋅ {(1, [1, 1]) − (0.8, [ , 1])} = (1, [1, 1])
2
𝑃3 = [0.4, 0.8] (13)
2.16
− (0.056, [ , 1]) = [0.568, 0.944]
0.28
Substituting the values in (11), (12), and (13) into (7) to
(10), the probability interval range of the top event of the fault 1
𝑃𝑇3 = (1, [1, 1]) − {(1, [1, 1]) − (0.3, [ , 1])}
tree can be obtained as follows: 3
1 1
𝑃𝑇1 = [1, 1] − [0.7, 0.9] ⋅ [0.4, 0.8] ⋅ [0.2, 0.6] = [1, 1] ⋅ {(1, [1, 1]) − (0.6, [ , 1]) − (0.8, [ , 1]) (23)
3 2
(14)
− [0.056, 0.432] = [0.568, 0.944] 1 1
+ (0.6, [ , 1]) ⋅ (0.8, [ , 1])} = [0.568, 0.944]
3 2
𝑃𝑇2 = [1, 1] − ([1, 1] − [0.1, 0.3] − [0.2, 0.6]
1 1 1
𝑃𝑇4 = (0.3, [ , 1]) + (0.6, [ , 1]) + (0.8, [ , 1])
+ [0.1, 0.3] ⋅ [0.2, 0.6]) ⋅ ([1, 1] − [0.4, 0.8]) = [1, 1] (15) 3 3 2
− [0.12, 0.88] ⋅ [0.2, 0.6] = [0.472, 0.976] 1 1 1
− (0.3, [ , 1]) ⋅ (0.6, [ , 1]) − (0.3, [ , 1])
3 3 3
𝑃𝑇3 = [1, 1] − ([1, 1] − [0.1, 0.3]) ⋅ ([1, 1] − [0.2, 0.6]
1 1 1 (24)
⋅ (0.8, [ , 1]) − (0.6, [ , 1]) ⋅ (0.8, [ , 1])
− [0.4, 0.8] + [0.2, 0.6] ⋅ [0.4, 0.8]) = [1, 1] (16) 2 3 2
− [0.7, 0.9] ⋅ [−0.32, 0.88] = [0.208, 1.288] 1 1 1
+ (0.3, [ , 1]) ⋅ (0.6, [ , 1]) ⋅ (0.8, [ , 1])
3 3 2
𝑃𝑇4 = [0.1, 0.3] + [0.2, 0.6] + [0.4, 0.8] − [0.1, 0.3]
= [0.568, 0.944]
⋅ [0.2, 0.6] − [0.1, 0.3] ⋅ [0.4, 0.8] − [0.2, 0.6]
(17) For the four equivalent forms of the failure function of the
⋅ [0.4, 0.8] + [0.1, 0.3] ⋅ [0.2, 0.6] ⋅ [0.4, 0.8] fault tree composed of three OR gate bottom events shown
in Figure 1, four different failure probability interval values
= [−0.192, 1.704] of the top event are obtained using the traditional interval
4 Complexity

T
For the AND gate operator, the failure probability inter-
vals of the three bottom events can be explained in terms of
universal grey numbers as follows:
1
𝑋1 = (0.3, [ , 1]) (30)
3
X1 X2 X3
1
𝑋2 = (0.6, [ , 1]) (31)
Figure 2: AND gate operator with three bottom events. 3
1
𝑋3 = (0.8, [ , 1]) (32)
2
The probability interval value of the top event of the fault
computation: [0.568, 0.944], [0.472, 0.976], [0.208, 1.288], tree as determined by the universal grey operation is thus
and [-0.192, 1.704]. Because (7) has a physical meaning and is
not simplified, [0.568, 0.944] is the correct result. Although 1 1 1
𝑋𝑇 = (0.3, [ , 1]) ⋅ (0.6, [ , 1]) ⋅ (0.8, [ , 1])
(8) to (10) are equivalent to (7) in form, different orders 3 3 2 (33)
of interval computation can result in different expansion
degrees of the failure probability with respect to the top event = (0.144, [0.0556, 1]) = [0.008, 0.144]
in the fault tree. Because the failure function of the three bottom events
The failure probability interval of the top event calculated in the AND gate operator has only one form, the traditional
using the four equivalent forms is [0.568, 0.944] as deter- interval analysis and the proposed grey number interval anal-
mined through the universal grey operation, the same value ysis will provide the same solution for the failure probability
obtained using (7), but without the enlargement or reduction of the top event. For the AND gate, the traditional interval
by different degrees when a traditional interval operation is arithmetic is equivalent to the universal grey operation. In
applied using different orders of operation. This indicates actual engineering, however, complex system fault trees are
that an interval analysis combined with grey operation can generally composed of multiple AND gates and OR gates.
overcome the drawbacks of traditional interval operations. Therefore, in order to demonstrate the effectiveness of the
proposed universal grey operation, it is necessary to perform
an interval analysis of a complex system fault tree that is
2.4. Interval Analysis of Fault Trees Based on the Universal
appropriate for universal grey operation.
Grey Operator for an AND Gate Operator. A fault tree interval
analysis based on the universal grey operator is illustrated by
using an AND gate operator with three bottom events as an 3. Interval Analysis of Landing Gear
example. The fault tree is shown in Figure 2. Retraction System Fault Tree Based on
The failure function of the top event related to the bottom Universal Grey Operation
events is given by
3.1. Fault Tree Model of Landing Gear Retraction System. An
aircraft landing gear retraction system mainly consists of
𝑋𝑇 = 𝑋1 ⋅ 𝑋2 ⋅ 𝑋3 (25) hydraulic retractable cylinders, shock struts, foldable rear
struts, two rotation shafts fixed to the airframe, and other
corresponding attachments. When an aircraft extends its
The failure probabilities of the three bottom events landing gear, the lower lock opens first, then hydraulic oil
are indicated by their interval numbers with the following is injected into the actuating cylinder, causing its piston rod
interval values: to extend outward, pushing the landing gear shock strut to
rotate it about the front rotation shaft. When the landing
gear is in place, the upper lock is closed and the injection of
𝑋1 = [0.1, 0.3] (26) hydraulic oil stops. When an aircraft retracts its landing gear,
the upper lock opens first, then the hydraulic oil is injected
𝑋2 = [0.2, 0.6] (27) into the actuating cylinder causing its piston rod to retract
inward, pulling the landing gear shock strut to rotate it about
𝑋3 = [0.4, 0.8] (28) the front rotation shaft. Once the landing gear is stowed, the
lower lock is closed and the injection of hydraulic oil stops
[34].
The failure probability of the top event can then be In this study, the system fault tree was constructed
calculated according to the traditional interval operation according to the working principle, composed structure, and
using: fault classification of the landing gear retraction system, as
shown in Figure 3.
This paper reviews the failure probability interval range of
𝑋𝑇 = [0.1, 0.3] ⋅ [0.2, 0.6] ⋅ [0.4, 0.8] = [0.008, 0.144] (29) each bottom event in the landing gear retraction system fault
Complexity 5

M1 M2 M3

+ + +

X1 M4 M5 M6 X10 X11 X12 X13 X14 M7 X20

X18 X19

+ + + +

X2 X3 X4 X5 X6 X7 X8 X9 X15 X16 X17

Figure 3: Fault tree of an aircraft landing gear retraction system, where T is a landing gear retraction system fault; M1 is an actuating cylinder
fault; M2 is a control system fault; M3 is a hydraulic system fault; M4 is a piston rod fault; M5 is a sealing ring fault; M6 is an end eyelet fault;
M7 is a hydraulic pump fault; X1 is a cylinder fracture; X2 is a seized piston rod; X3 is a loose piston rod; X4 is a fractured piston rod; X5 is
a mechanically damaged sealing ring; X6 is an aging and cracking sealing ring; X7 is a chemically corroded sealing ring; X8 is an eyelet bolt
fracture; X9 is an eyelet rotation; X10 is the failure of the pressure switch; X11 is the failure of the electromagnetic switch; X12 is the failure of
the safety valve; X13 is oil leakage from the hydraulic line; X14 is an air lock fault; X15 is a fracture of the pump plunger piston spring; X16 is
oil leakage from the pump rotary joint; X17 is a damaged pump motor; X18 is oil contamination; X19 is a broken hard pipe; X20 is a break in
the accumulator.

Table 1: Interval of basic event failure probability in the landing gear retraction system fault tree.

Fault code Interval of basic event failure probability Fault code Interval of basic event failure probability
X1 [2.02 × 10−5 , 6.40 × 10−5 ] X11 [2.14 × 10−4 , 8.09 × 10−4 ]
X2 [2.62 × 10−5 , 9.29 × 10−5 ] X12 [4.95 × 10−4 , 1.44 × 10−3 ]
X3 [6.42 × 10−5 , 1.40 × 10−4 ] X13 [2.62 × 10−3 , 7.20 × 10−3 ]
X4 [2.10 × 10−6 , 8.85 × 10−6 ] X14 [2.62 × 10−4 , 8.25 × 10−4 ]
X5 [4.60 × 10−4 , 2.45 × 10−3 ] X15 [4.60 × 10−5 , 1.85 × 10−4 ]
X6 [2.62 × 10−4 , 8.29 × 10−4 ] X16 [2.44 × 10−4 , 7.52 × 10−4 ]
X7 [2.44 × 10−4 , 9.72 × 10−4 ] X17 [2.45 × 10−3 , 7.72 × 10−3 ]
X8 [4.60 × 10−4 , 1.95 × 10−3 ] X18 [2.79 × 10−3 , 9.85 × 10−3 ]
X9 [2.24 × 10−4 , 7.29 × 10−4 ] X19 [2.12 × 10−5 , 8.29 × 10−5 ]
X10 [1.62 × 10−5 , 8.29 × 10−5 ] X20 [4.42 × 10−5 , 1.70 × 10−4 ]

tree, shown in Table 1. By referring to the logical relationship 3.2. Failure Probability of Top Event of the System Fault
OR gates in the fault tree, the probabilistic structure function Tree under Universal Grey Operation. Depending on the
model of the fault tree can be obtained as conversion rule of the interval number and universal grey
number, the failure probabilities of the bottom events of
𝑃𝑇 = 1 − (1 − 𝑃𝑀1 ) (1 − 𝑃𝑀2 ) (1 − 𝑃𝑀3 ) (34) the landing gear retraction system, expressed using universal
grey numbers, are given in Table 2.
𝑃𝑀1 = 1 − (1 − 𝑃𝑋1 ) (1 − 𝑃𝑀4 ) (1 − 𝑃𝑀5 ) (1 − 𝑃𝑀6 ) The universal grey operation process is first conducted as
follows on the lowest level of gates:
𝑃𝑀2 = 1 − (1 − 𝑃𝑋10 ) (1 − 𝑃𝑋11 ) (1 − 𝑃𝑋12 )
(35)
𝑃𝑀3 = 1 − (1 − 𝑃𝑋13 ) (1 − 𝑃𝑋14 ) (1 − 𝑃𝑀7 ) (1 − 𝑃𝑋18 ) 𝑃𝑀4 = 1 − (1 − 𝑃𝑋2 ) (1 − 𝑃𝑋3 ) (1 − 𝑃𝑋4 )
⋅ (1 − 𝑃𝑋19 ) (1 − 𝑃𝑋20 ) = (1, [1, 1])

𝑃𝑀4 = 1 − (1 − 𝑃𝑋2 ) (1 − 𝑃𝑋3 ) (1 − 𝑃𝑋4 ) − {(1, [1, 1]) − (9.29 × 10−5 , [0.282024, 1])}
(37)
𝑃𝑀5 = 1 − (1 − 𝑃𝑋5 ) (1 − 𝑃𝑋6 ) (1 − 𝑃𝑋7 ) ⋅ {(1, [1, 1]) − (1.40 × 10−4 , [0.458571, 1])}
(36)
𝑃𝑀6 = 1 − (1 − 𝑃𝑋8 ) (1 − 𝑃𝑋9 ) ⋅ {(1, [1, 1]) − (8.85 × 10−6 , [0.237288, 1])}
𝑃𝑀7 = 1 − (1 − 𝑃𝑋15 ) (1 − 𝑃𝑋16 ) (1 − 𝑃𝑋17 ) = (2.4 × 10−4 , [0.3793, 1])
6 Complexity

Table 2: Universal grey representation of basic event failure probability for landing gear retraction system fault tree.

Fault code Universal grey representation of failure probability Fault code Universal grey representation of failure probability
X1 (6.40 × 10−5 , [0.315625, 1]) X11 (8.09 × 10−4 , [0.264524, 1])
X2 (9.29 × 10−5 , [0.282024, 1]) X12 (1.44 × 10−3 , [0.343750, 1])
X3 (1.40 × 10−4 , [0.458571, 1]) X13 (7.20 × 10−3 , [0.363889, 1])
X4 (8.85 × 10−6 , [0.237288, 1]) X14 (8.25 × 10−4 , [0.317576, 1])
X5 (2.45 × 10−3 , [0.187755, 1]) X15 (1.85 × 10−4 , [0.248649, 1])
X6 (8.29 × 10−4 , [0.316043, 1]) X16 (7.52 × 10−4 , [0.324468, 1])
X7 (9.72 × 10−4 , [0.251029, 1]) X17 (7.72 × 10−3 , [0.317358, 1])
X8 (1.95 × 10−3 , [0.235897, 1]) X18 (9.85 × 10−3 , [0.283249, 1])
X9 (7.29 × 10−4 , [0.307270, 1]) X19 (8.29 × 10−5 , [0.255730, 1])
X10 (8.29 × 10−5 , [0.195416, 1]) X20 (1.70 × 10−4 , [0.260000, 1])

𝑃𝑀5 = 1 − (1 − 𝑃𝑋5 ) (1 − 𝑃𝑋6 ) (1 − 𝑃𝑋7 ) ⋅ {(1, [1, 1]) − (4.246 × 10−3 , [0.227505, 1])}
= (1, [1, 1]) ⋅ {(1, [1, 1]) − (2.678 × 10−3 , [0.2555464, 1])}
− {(1, [1, 1]) − (2.45 × 10−3 , [0.187755, 1])}
= (7.2145 × 10−3 , [0.2431464, 1])
(38)
⋅ {(1, [1, 1]) − (8.29 × 10−4 , [0.316043, 1])} (41)

⋅ {(1, [1, 1]) − (9.72 × 10−4 , [0.251029, 1])} 𝑃𝑀2 = 1 − (1 − 𝑃𝑋10 ) (1 − 𝑃𝑋11 ) (1 − 𝑃𝑋12 )
= (1, [1, 1])
= (4.246 × 10−3 , [0.227505, 1])
− {(1, [1, 1]) − (8.29 × 10−5 , [0.195416, 1])}
𝑃𝑀6 = 1 − (1 − 𝑃𝑋8 ) (1 − 𝑃𝑋9 )
(42)
= (1, [1, 1]) ⋅ {(1, [1, 1]) − (8.09 × 10−4 , [0.264524, 1])}

− {(1, [1, 1]) − (1.95 × 10−3 , [0.235897, 1])} (39) ⋅ {(1, [1, 1]) − (1.44 × 10−3 , [0.343750, 1])}

⋅ {(1, [1, 1]) − (7.29 × 10−4 , [0.307270, 1])} = (2.33 × 10−3 , [0.310623, 1])

= (2.678 × 10−3 , [0.2555464, 1]) 𝑃𝑀3 = 1 − (1 − 𝑃𝑋13 ) (1 − 𝑃𝑋14 ) (1 − 𝑃𝑀7 ) (1 − 𝑃𝑋18 )

𝑃𝑀7 = 1 − (1 − 𝑃𝑋15 ) (1 − 𝑃𝑋16 ) (1 − 𝑃𝑋17 ) ⋅ (1 − 𝑃𝑋19 ) (1 − 𝑃𝑋20 ) = (1, [1, 1])

= (1, [1, 1]) − {(1, [1, 1]) − (7.20 × 10−3 , [0.363889, 1])}

− {(1, [1, 1]) − (1.85 × 10−4 , [0.248649, 1])} ⋅ {(1, [1, 1]) − (8.25 × 10−4 , [0.317576, 1])}
(40)
⋅ {(1, [1, 1]) − (7.52 × 10−4 , [0.324468, 1])} ⋅ {(1, [1, 1]) − (8.65 × 10−3 , [0.3167134, 1])} (43)

⋅ {(1, [1, 1]) − (7.72 × 10−3 , [0.317358, 1])} ⋅ {(1, [1, 1]) − (9.85 × 10−3 , [0.283249, 1])}

= (8.65 × 10−3 , [0.3167134, 1]) ⋅ {(1, [1, 1]) − (8.29 × 10−5 , [0.255730, 1])}

⋅ {(1, [1, 1]) − (1.70 × 10−4 , [0.260000, 1])}

Allowing the universal grey operation to be conducted as
follows on the next highest level of gates, = (2.65324 × 10−2 , [0.318568, 1])

𝑃𝑀1 = 1 − (1 − 𝑃𝑋1 ) (1 − 𝑃𝑀4 ) (1 − 𝑃𝑀5 ) (1 − 𝑃𝑀6 ) Then, for the top event,

= (1, [1, 1]) 𝑃𝑇 = 1 − (1 − 𝑃𝑀1 ) (1 − 𝑃𝑀2 ) (1 − 𝑃𝑀3 )

− {(1, [1, 1]) − (6.40 × 10−5 , [0.315625, 1])} = (1, [1, 1])

⋅ {(1, [1, 1]) − (2.4 × 10−4 , [0.3793, 1])} − {(1, [1, 1]) − (7.2145 × 10−3 , [0.2431464, 1])}
Complexity 7

⋅ {(1, [1, 1]) − (2.33 × 10−3 , [0.310623, 1])} [4] P. Liu, L. Yang, Z. Gao, S. Li, and Y. Gao, “Fault tree analysis
combined with quantitative analysis for high-speed railway
⋅ {(1, [1, 1]) − (2.65324 × 10−2 , [0.318568, 1])} accidents,” Safety Science, vol. 79, pp. 344–357, 2015.
[5] C. Cao, M. Li, Y. Li, and Y. Sun, “Intelligent fault diagnosis of
= (0.0358, [0.3044832, 1]) = [0.0109, 0.0358] hot die forging press based on binary decision diagram and
fault tree analysis,” Procedia Manufacturing, vol. 15, pp. 459–
(44)
466, 2018.
indicating that the failure probability interval range of the [6] Y. Deng, H. Wang, and B. Guo, “BDD algorithms based on
modularization for fault tree analysis,” Progress in Nuclear
aircraft landing gear retraction system varies between 0.0109
Energy, vol. 85, pp. 192–199, 2015.
and 0.0358.
[7] S. Kabir, “An overview of fault tree analysis and its application
in model based dependability analysis,” Expert Systems with
4. Conclusions Applications, vol. 77, pp. 114–135, 2017.
[8] L. Cai, L. Zhang, S. Yang, and L. Wang, “Reliability assessment
This paper proposes a new method for system fault tree and analysis of large aircraft power distribution systems,”
analysis that overcomes the shortcomings of existing interval Hangkong Xuebao/Acta Aeronautica et Astronautica Sinica, vol.
operation methods based on the conversion rules between 32, no. 8, pp. 1488–1496, 2011.
the interval number and the universal grey number and [9] L.-D. Jia, J.-A. Wang, Z.-T. Yang, and G.-Y. Tu, “Analysis and
describes the failure probability interval range of the bottom modeling of leak factors of spaceship leak-testing room based
events of a fault tree using the universal numbers. Based on FTA method,” Yuhang Xuebao/Journal of Astronautics, vol.
on the logical relationship between the AND gates and OR 33, no. 6, pp. 843–848, 2012.
gates of a fault tree and combined with the four arithmetic [10] X.-J. Zhang, H.-T. Zhao, Q. Miao, W. Zhang, and H.-Z. Huang,
operations of the universal grey number, a more accurate “Reliability analysis of satellite system based on dynamic fault
system reliability calculation result is achieved. The proposed tree,” Yuhang Xuebao/Journal of Astronautics, vol. 30, no. 3, pp.
method is demonstrated to be advantageous for interval 1249–1254, 2009.
operation and is shown to overcome the drawbacks of [11] H. M. Dong and Z. M. Dong, “Analysis and design of fault
enlarged or reduced solution intervals when using different tree for integrated dry route furnace in nuclear fuel fabrication
orders of interval operation. process,” Nuclear Science and Engineering, vol. 35, no. 4, pp. 688–
693, 2015.
[12] N. J. Hu, “Fault diagnosis of servo system assembly based
Data Availability on FTA for certain type antiaircraft gun,” Sichuan Ordnance
Journal, vol. 32, no. 4, pp. 40–43, 2011.
The data used to support the findings of this study are [13] L. Zhang, Y. He, W. Zhai, J. Zhao, and F. Zhang, “The Fault
included within the article. Tree Analysis Of aircraft Hydraulic Brake System Based on
Hyper-ellipsoidal Model,” Jixie Qiangdu/Journal of Mechanical
Conflicts of Interest Strength, vol. 39, no. 4, pp. 842–847, 2017.
[14] C. Y. Yao, J. Lu, D. N. Chen et al., “Convex model T-S fault
The authors declare that they have no conflicts of interest. tree and importance analysis methods,” Journal of Mechanical
Engineering, vol. 51, no. 24, pp. 184–192, 2015.
[15] F. Piadeh, M. Ahmadi, and K. Behzadian, “Reliability assess-
Acknowledgments ment for hybrid systems of advanced treatment units of indus-
trial wastewater reuse using combined event tree and fuzzy fault
The authors gratefully acknowledge the support of China
tree analyses,” Journal of Cleaner Production, vol. 201, pp. 958–
Scholarship Council, Top International University Visiting 973, 2018.
Program for Outstanding Young scholars of Northwest-
[16] A. Cheliyan and S. Bhattacharyya, “Fuzzy fault tree analysis of
ern Polytechnical University, the Fundamental Research oil and gas leakage in subsea production systems,” Journal of
Fund for the Central Universities (NPU-FFR-3102015BJ(II) Ocean Engineering and Science, vol. 3, no. 1, pp. 38–48, 2018.
JL04), and Natural Science Foundation of Shaanxi Province
[17] J. H. Purba, D. T. Sony Tjahyani, A. S. Ekariansyah, and H.
(2016JQ5109). Tjahjono, “Fuzzy probability based fault tree analysis to prop-
agate and quantify epistemic uncertainty,” Annals of Nuclear
References Energy, vol. 85, pp. 1189–1199, 2015.
[18] A. Mottahedi and M. Ataei, “Fuzzy fault tree analysis for coal
[1] S. Lin, Y. Wang, and L. Jia, “System Reliability Assessment Based burst occurrence probability in underground coal mining,”
on Failure Propagation Processes,” Complexity, vol. 2018, Article Tunnelling and Underground Space Technology, vol. 83, pp. 165–
ID 9502953, 19 pages, 2018. 174, 2019.
[2] F. Sihombing and M. Torbol, “Parallel fault tree analysis for [19] Y. Ding and A. Lisnianski, “Fuzzy universal generating func-
accurate reliability of complex systems,” Structural Safety, vol. tions for multi-state system reliability assessment,” Fuzzy Sets
72, pp. 41–53, 2018. and Systems, vol. 159, no. 3, pp. 307–324, 2008.
[3] L. Giraud and B. Galy, “Fault tree analysis and risk mitigation [20] Y.-F. Li, Y. Ding, and E. Zio, “Random fuzzy extension of
strategies for mine hoists,” Safety Science, vol. 110, pp. 222–234, the universal generating function approach for the reliability
2018. assessment of multi-state systems under aleatory and epistemic
8 Complexity

uncertainties,” IEEE Transactions on Reliability, vol. 63, no. 1, pp.

13–25, 2014.
[21] Y. Liu and H.-Z. Huang, “Reliability assessment for fuzzy multi-
state systems,” International Journal of Systems Science, vol. 41,
no. 4, pp. 365–379, 2010.
[22] J. Liu, H. Liu, C. Jiang, X. Han, D. Q. Zhang, and Y. F. Hu,
“A new measurement for structural uncertainty propagation
based on pseudo-probability distribution,” Applied Mathemati-
cal Modelling: Simulation and Computation for Engineering and
Environmental Systems, vol. 63, pp. 744–760, 2018.
[23] L. Cao, J. Liu, X. Han, C. Jiang, and Q. Liu, “An efficient
evidence-based reliability analysis method via piecewise hyper-
plane approximation of limit state function,” Structural and
Multidisciplinary Optimization, vol. 58, no. 1, pp. 201–213, 2018.
[24] F. Zhang, K. Luo, W. Zhai, S. Tan, and Y. Wang, “Non-
probabilistic parameter sensitivity analysis for structures based
on ellipsoidal model,” Advances in Mechanical Engineering, vol.
10, no. 8, pp. 1–7, 2018.
[25] C. Jiang, Q. F. Zhang, X. Han, J. Liu, and D. A. Hu, “Multidimen-
sional parallelepiped model—a new type of non-probabilistic
convex model for structural uncertainty analysis,” International
Journal for Numerical Methods in Engineering, vol. 103, no. 1, pp.
31–59, 2015.
[26] C. Jiang, B. Y. Ni, X. Han, and Y. R. Tao, “Non-probabilistic
convex model process: a new method of time-variant uncer-
tainty analysis and its application to structural dynamic reli-
ability problems,” Computer Methods Applied Mechanics and
Engineering, vol. 268, pp. 656–676, 2014.
[27] C. Jiang, R. G. Bi, G. Y. Lu, and X. Han, “Structural reliabil-
ity analysis using non-probabilistic convex model,” Computer
Methods Applied Mechanics and Engineering, vol. 254, pp. 83–
98, 2013.
[28] J. Quan, B. Zeng, and D. Liu, “Green Supplier Selection for
Process Industries Using Weighted Grey Incidence Decision
Model,” Complexity, vol. 2018, Article ID 4631670, 12 pages,
2018.
[29] J. Liu, Q. X. Qing, Q. C. Zhang et al., “Solving of the structural
fuzzy reliability analysis method with universal grey numbers,”
Applied Mathematics and Mechanics, vol. 36, no. 3, pp. 285–293,
2015.
[30] D. Luo, “Risk evaluation of ice-jam disasters using gray systems
theory: The case of Ningxia-Inner Mongolia reaches of the
Yellow River,” Natural Hazards, vol. 71, no. 3, pp. 1419–1431, 2014.
[31] H. Jin, J. Chen, H. Ma, and F. You, “Method for calculation of
Rayleigh quotient with universal grey mathematics,” Zhendong
Ceshi Yu Zhenduan/Journal of Vibration, Measurement and
Diagnosis, vol. 33, no. 3, pp. 416–420, 2013.
[32] J. Liu, M. Wang, and Q. X. Qing, “A new structural non-
probabilistic reliability analysis method,” Acta Mechanica Solida
Sinica, vol. 35, no. 4, pp. 378–383, 2014.
[33] H. Jin, J. Chen, H. Cao, and Y. Xu, “Universal grey analysis
for natural frequencies of uncertain chain structures system,”
Journal of Vibration, Measurement and Diagnosis, vol. 36, no. 2,
pp. 252–256, 2016.
[34] H. Yong, Reliability Design and Analysis of Aircraft Landing Gear
Subsystem, Northwestern Polytechnical University, Xi’an, 2013.