You are on page 1of 7

Double Layer Fuzzy enhanced Failure Mode and

Effects Analysis Method for Automatic Train


Operation System

Wei ShangGuan
Ziqi Wang
School of Electronic and Information Engineering
School of Electronic and Information Engineering the State Key Laboratory of Rail Traffic Control
Beijing Jiaotong University Beijing Jiaotong University
Beijing, PR China Beijing, PR China
20125106@bjtu.edu.cn wshg@bjtu.edu.cn

Cong Peng Hongyu Song


School of Electronic and Information Engineering School of Electronic and Information Engineering
Beijing Jiaotong University Beijing Jiaotong University
Beijing, PR China Beijing, PR China
19111064@bjtu.edu.cn Hongyusong@bjtu.edu.cn

Abstract—Automatic Train Operation (ATO) is a vital part of the internal hierarchy and high maintenance cost. Without affecting
rail traffic signal system. Failure Mode Effects and Criticality the safety and efficiency of train operation, PHM is usually used
analysis (FMECA) is the first step of Prognostic and Health to know the health status of ATO system in real time and make
Management (PHM). In this paper, this method based on double optimal maintenance strategies. And FMECA is the first step of
layer fuzzy enhanced failure mode and effects analysis (FMEA) is health management. But in the traditional FMECA method, the
proposed. Firstly, ATO failure analyzer is applied to classify and ranking of failures is based on the Risk Priority Number (RPN),
score ATO failure causes. Secondly, expert scoring mechanism is which is calculated by product the three factors S, O and D [2]. It
used for fuzzy aggregation of multiple scoring opinions. Finally,
ignores some important hidden problems. To remedy the
Fuzzy Risk Priority Number (FRPN) is obtained by fuzzy risk
traditional FMECA insufficiency, significant efforts have been
estimator which is employed for fuzzy aggregation on Severity (S),
Occurrence (O) and Detection (D) get in the second step. The result
made by combine the FMECA with other methods [3-5]. N. G.
shows that this proposed method can find out the potential failure Mutlu and S. Alyuntas [3] proposed an approach based on the
causes which are same Risk Priority Number (RPN) value in integration of FMEA with the FTA method and Belief in Fuzzy
traditional FMECA. The reliability of ATO system can be Probability Estimations of Time (BIFPET) algorithm for
guaranteed. finishing process in the fabric dyeing department of a textile
company. C. FenChi, D.Sigmund and M.O. Astardi [4] applied
Keywords-component; Automatic train operation; Double layer the root cause analysis, the functional block diagram, and failure
fuzzy; Failure mode and effects analysis; modes and effects analysis (FMEA) for passenger vehicle recalls.
N. Chanamool and T.Naenna[5] applied the Fuzzy failure mode
I. INTRODUCTION and effect analysis (FMEA) to assess the failures in the
With the acceleration of urbanization, urban rail transit has emergency department. Because the subsystem structure and
entered a period of rapid development. Up to now, a total of 45 working mechanism of ATO are typically characterized by
cities in China have opened urban rail transit, with 244 operating complexity, multiple targets, delay and nonlinear dynamic
lines and a total length of 7,969.7 kilometers. ATO mainly system. In the context, this paper describes a fuzzy aggregation
realizes the automatic control of train traction and service opinion and a fuzzy risk priority number assessment for ATO
braking system, as well as generating the command of the failure cause based on FMEA and fuzzy logic method. The study
automatic opening/closing of train door, platform screen door [1]. is compared by the traditional method and the fuzzy ones
ATO system is composed of components with high reliability through the Pareto chart.
and strong anti-interference ability through logic combination This paper is structures as follows. In Section 2, a brief
and bus structure, and its equipment composition is shown in Fig. review of FMEA and Fuzzy logic approach are given. Section 3
1. presents the methodology of research and double layer fuzzy
The ATO system operation are increasingly becoming the enhanced FMEA frame. Section 4 is completed by a discussion
focus of the public due to its high security level, complex of the results and a comparison of the proposed double layer
Corresponding authors: Wei ShangGuan (wshg@bjtu.edu.cn)
2021 Global Reliability and Prognostics and Health Management
(PHM-Nanjing)
fuzzy enhanced FMEA with traditional FMECA. The results are TABLE II. CRITERIA DEFINED FOR OCCURRENCE SCORING IN FMEA
verified by Pareto chart. Finally, some conclusions are provided
Rating Occurrence Probability Failure probability
in Section 5.
10 Extremely high ≥ 1/2

9 Very high 1/3

8 Repeated failures 1/8

7 High 1/20

6 Moderately high 1/80

5 Moderate 1/400

4 Relatively low 1/2000

3 Low 1/15000
Figure 1. Equipment constitution of ATO system 2 Remote 1/150000

II. PRELIMINARIES 1 Nearly impossible ≤ 1/1500000

A. Failure mode and effects analysis


TABLE III. CRITERIA DEFINED FOR DETECTION SCORING IN FMEA
FMEA is a widely used engineering technology, which is
used to define, identify and eliminate known or potential faults, Rating Detection definition
problems and errors, etc. FMEA is a combination of failure
10 Absolute uncertainty possibility
mode analysis and failure impact analysis, enabling designers or
operators to reduce or even eliminate various possibilities of 9 Very remote possibility
research objects based on existing data. In the 1960s, FMEA was
8 Remote possibility
first put forward by NASA and applied in many projects,
including Apollo, Viking, Voyager, Magellan, Galileo and 7 Very low possibility
Skylab. [6]Since then, it has been widely used as a powerful tool
6 Low possibility
for safe and reliable analysis of products and processes in a wide
range of industries, especially in aerospace, nuclear energy and 5 Moderately possibility
automobile industries.[7] After the analysis object is determined,
4 Moderately high possibility
FMEA firstly identifies the failure effect, failure cause and
detectability evaluation step by step, and obtains three factors, 3 High possibility
S(Severity), O(Occurrence) and D(Detection), which are scored
2 Very high possibility
10-points scale (defined in TABLE Ⅰ-Ⅲ) depending on
criticality nature[8]. The “S” index takes into account the 1 Almost certain possibility
criticality of the failure effect on the ATO system. The “O” index
estimates the frequency of failure while the “D” index estimates B. Fuzzy Logic Approach
the probability of detecting the failure cause. RPN as the Fuzzy logic refers to a description system whose model is
traditional FMEA approach is calculated per the Eq. (1) [9-10]: unknown or uncertain, which uses fuzzy sets and fuzzy rules to
reason, express transition boundaries or qualitative knowledge
𝑅𝑃𝑁 = 𝑆 × 𝑂 × 𝐷 () and experience, and conduct fuzzy comprehensive evaluation,
so as to solve the problem of fuzzy information of rules which is
TABLE I. CRITERIA DEFINED FOR SEVERITY SCORING IN FMEA difficult to be handled by conventional methods. Membership
Rating Failure Effect
function is an important concept in fuzzy logic. It is a curve with
arbitrary shape, which maps the specific input elements to an
10 Hazardous without warning appropriate membership degree (between zero to 1). The
9 Hazardous with warning membership functions used in this paper include Triangular,
Gaussian, Z-shaped and S-shaped membership functions.
8 Hazardous without warning
Triangular fuzzy membership has the following membership
7 Hazardous with warning function:
6 Extreme effect
𝑥−𝑎
5 Very high effect 𝑎≤𝑥≤𝑏
𝑏−𝑎
4 High effect 𝜇(𝑥) = { 𝑑−𝑥 𝑏≤𝑥≤𝑑 ()
𝑑−𝑏
3 Moderate effect 0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
2 Low effect
Gaussian fuzzy membership has the following membership
1 Very effect
function:

2021 Global Reliability and Prognostics and Health Management


(PHM-Nanjing)
2 ⁄2𝑒 2
𝜇(𝑥, 𝑐, 𝜎) = 𝑒 −(𝑥−𝑐) (3)

where σ is the standard deviation and c is the mean of the


Gaussian function.
Z-shared fuzzy membership has the following membership
function:

1 𝑥≤𝑎
𝑥−𝑎 2 𝑎+𝑏
1− 2( ) 𝑎≤𝑥≤
𝜇(𝑥) = 𝑏−𝑎 2 (4)
𝑥−𝑏 2 𝑎+𝑏
2( ) ≤𝑥≤𝑏
𝑏−𝑎 2
{ 0 𝑥≥𝑏
S-shared fuzzy membership has the following membership
function:

0 𝑥≤𝑎
𝑥−𝑏 2 𝑎+𝑏
2( ) 𝑎≤𝑥≤
𝑏−𝑎 2
𝜇(𝑥) = (5)
𝑥−𝑎 2 𝑎+𝑏
1− 2( ) ≤𝑥≤𝑏
𝑏−𝑎 2
{ 1 𝑥≥𝑏
Fuzzy logic can deal with accurate and inaccurate
information in a consistent way. it can combine S, O and D in
more practical way. And the risk assessment function may vary
from system to system [11-13].
Figure 2. Double layer fuzzy enhanced FMEA frame
III. METHODOLOGY
In this paper, we propose the method based on double layer TABLE IV. FMEA FOR ATO SYSTEM
fuzzy enhanced FMEA. It divides the evaluation process of ATO Failure
system into ATO failure analyzer, expert scoring mechanism Function Failure modes Failure cause cause
and ATO failure fuzzy risk estimator as shown in the Fig. 2. number
Supporting an expert team is consisted of m member {TM𝑚 (m= -Damaged in the power FM1
interface
1, 2, ..., M)} which assess failure modes {FM𝑛 (n= 1, 2, ..., N)} -Inverter overload FM2
with three risk factors (S, O&D). Power failure
-Inverter temperature too
A. ATO failure analyzer FM3
high
Although ATO system used by domestic urban rail transit
operators are different in structure, the basic components of -Battery low voltage FM4
ATO are the same, which are composed of on-board equipment -Communication with HMI FM5
and trackside equipment. ATO is a subsystem responsible for Communication
failed
automatic train operation. It controls train operation according template failure
-Loss ATP signal FM6
to the speed-distant curve and ATO curve calculated by ATO
algorithm. ATO system realizes optimal control and the best -electromagnetic
ATO FM7
interference
running state of the train by simulating the driver and it follows Input interface
the closed-loop control in principle. Its specific control process failure
-Worn input interface FM8
is as follows: the speed measuring motor and radar transmit the
information of train position to ATO through ATP, the input of -Fault of tachogenerator FM9
the reference position comes from the position information of
the transponder, and ATO provides the set value data for output -Memory burned out FM10
Central
to the traction and braking system. Under this context, FMEA processing board
-corrupted in the network FM11
was used for it to carry out quantitative analysis (TABLE Ⅳ). failure
adapted plug-in
-Traction fault FM12
B. Expert scoring mechanism Output interface
In this mechanism, the method based on fuzzy rules is fault -Display crashed FM13
adopted, and a logic relationship is established, in which
multiple expert evaluations are taken as inputs and a fuzzy
numerical value is output. The inputs and the output factors are

2021 Global Reliability and Prognostics and Health Management


(PHM-Nanjing)
represented by Triangular membership function. Using the fuzzy factors S, O and D, and changes the input membership degree
logic toolbox in MATLAB, according to the fuzzy rules in from zero to 1.
TABLE Ⅴ, a two-input and one-output fuzzy logic with a
numerical range of 1-10 is established. According to the above 2) Membership functions for FRPN output
description, two expert evaluations can be summarized into one. In the study, the minimum value of FRPN isFRPNmin = Smin ×
With the increase of expert team members, the number of fuzzy Omin × Dmin = 1. And the maximum value is FRPNmax =
set methods will also increase, and the process follows formula Smax × Omax × Dmax = 1000 . Therefore, the value range of
(6). FRPN is [1,1000]. This varied range of FRPN is divided into
eight levels. The fuzzy linguistic terms of these eight levels are
1 𝑛=1 shown in TABLE Ⅵ. The linguistic terms of FRPN
𝑁={ () output:1=Low, 2=High Low, 3=Low Moderate, 4=Moderate,
𝑛−1 𝑛 ≠1
5=High Moderate, 6=Low High, 7=High, 8=Very High. The
where n is the number of members in the expert team and N is output factors are represent using Gaussian membership
the total number of fuzzy sets. For example, N will be equal to function.
2 when there are three members in the expert team according to
Fig. 3.

Figure 3. Fuzzy aggregation structure based on 3-member team of


experts

Each red dot represents the calculation of fuzzy set. As


shown in above figure, after two calculations, the fuzzy set
Figure 4. Membership function for input factors: (a) severity input
evaluation value is generated [14]. variable, (b) occurrence input variable, (c) detection input variable.

TABLE V. THE RULES OF MULTI-EXPERT EVALUATION


TABLE VI. CLASSIFICATION OF FRPN OUTPUT
Expert 1
Range Linguistic Term
10 9 8 7 6 5 4 3 2 1
Expert 2 0<RPN<20 Low
10 10 10 9 9 8 8 7 7 6 6
20<RPN<50 High Low
9 10 9 9 8 8 7 7 6 6 5
8 9 9 8 8 7 7 6 6 5 5 50<RPN<100 Low Moderate

7 9 8 8 7 7 6 6 5 5 4 100<RPN<150 Moderate
6 8 8 7 7 6 6 5 5 4 3 150<RPN<250 High Moderate
5 8 7 7 6 6 5 5 4 3 3
250<RPN<350 Low High
4 7 7 6 6 5 5 4 4 3 2
3 7 6 6 5 5 4 4 3 2 2 350<RPN<500 High

2 6 6 5 5 4 4 3 3 2 1 500<RPN<1000 Very High


1 6 5 5 4 4 3 3 2 1 1 3) Fuzzy rules
The number of rules can be determined as “Number of
C. ATO failure fuzzy risk estimator Severity linguistic terms × Number of Occurrence linguistic
1) Membership functions for input factors terms×Number of Detection linguistic terms” [16]. In the study,
In double layer fuzzy enhanced FMEA, three risk factors (S, three input factors S, O and D generated 64 rules. These rules
O&D) are inputs of membership function [15]. The risk factors contain four linguistic terms, which are expressed in the IF-
ranging from 1 to 10 are divided into 4 levels, and their THEN format, for example:
corresponding linguistic terms in double layer fuzzy enhanced
FMEA are: 1=Low,4=Moderate, 7=High, 10=Very High, as • Rule 1: IF Severity is “Low”, Occurrence is “Low” and
shown in Fig. 4. The inputs factors are represent using Gaussian, Detection is “Low” THEN FRPN is “Low”;
Z-Shaped and S-Shaped membership function. Though the • Rule 36: IF Severity is “Very High”, Occurrence is
fuzzy toolbox in MATLAB ensures the fuzzification of input “Low” and Detection is “Very High” THEN FRPN is
“High Low”;

2021 Global Reliability and Prognostics and Health Management


(PHM-Nanjing)
• Rule 64: IF Severity is “Very High”, Occurrence is TABLE VIII. OCCURRENCE VALUE OF FAILURE CAUSE
“Very High” and Detection is “Very High” THEN Failure cause number TM1 TM2 TM3 Aggregated O value
FRPN is “Very High”;
FM1 4 6 4 5
The 64 rules are input into the rule editor, and the linguistic
FM2 7 5 4 5
terms of the input factors S, O and D are associated with the
linguistic terms of the output factor FRPN. Figs.5 (a), (b) and(c) FM3 8 4 5 6
show three-dimensional surface graphs of the relationship FM4 5 6 4 5
between input factors S, O and D and output FRPN [17-18]. FM5 3 4 2 2
FM6 1 4 5 3
FM7 5 5 5 6
FM8 7 5 8 7
FM9 2 6 5 5.03
FM10 4 4 3 5
FM11 2 5 1 2
FM12 3 5 4 5
FM13 5 8 6 7

TABLE IX. DETECTION VALUE OF FAILURE CAUSE

Failure cause number TM1 TM2 TM3 Aggregated D value

FM1 1 2 1 1

Figure 5. Fuzzy risk graph:(a) S and D factors changes versus FRPN FM2 5 5 6 6
output, (b) O and D factors changes versus FRPN output, (c) S and O factors FM3 8 5 4 6
changes versus FRPN output.
FM4 1 3 2 2
IV. SIMULATION EXPERIMENT AND ANALYSIS FM5 8 3 7 7
In this study, three members, who are TM1, TM2 and TM3, FM6 5 6 4 5
are experts with rich experience in ATO system maintenance. FM7 4 5 7 7
According to TABLE Ⅴ, three experts scored 13 failure causes
FM8 8 5 5 6
in three aspect (S, O&D) and calculated the fuzzy aggregated
values as shown in TABLE Ⅶ-Ⅸ. FM9 2 5 3 3
FM10 1 5 4 3
TABLE Ⅹ shows the risk priority number of each failure
causes and its ranking using traditional FMECA and double FM11 2 2 7 5
layer fuzzy enhanced FMEA. The failure FM8 has a rank of 1 FM12 5 1 2 2
with highest FRPN value of 271.5. The rank 13 was assigned to FM13 5 5 4 5
the lowest FRPN value equal to 29.85 for the failure cause FM4.
As shown in TABLE Ⅹ, the RPN value of FM2 and FM6 is
TABLE VII. SEVERITY VALUE OF FAILURE CAUSE the same, which both is 120. But their FRPN value is different,
which is 95.99 and 96.14 respectively. This proves that
Failure cause number TM1 TM2 TM3 Aggregated S value
traditional FMECA doesn’t take into account the importance
FM1 3 5 3 4.03 degree of each input factor and different value of S, O and D can
FM2 5 4 2 4
result the same RPN value. As a result, resources are wasted
while some high-risk potential failure reasons are not noticed.
FM3 3 6 8 7
Pareto chart is a kind if chart that arranges the failures
FM4 1 3 2 2
according to their important degree, which is a graphical
FM5 8 3 7 7 embodiment of “The 80/20 Rule”, that is, 80% of the problems
FM6 7 7 8 8 are caused by 20% of causes. In Pareto chart, different categories
of failures are arranged in descending order of RPN value, and
FM7 1 2 8 5 the cumulative percentage graph is drawn in the same graph. The
FM8 8 7 7 8 key reasons of most problems are mainly found through Pareto
FM9 4 4 6 6
chart. It can be seen from TABLE X that there are 12 categories
of RPN ranging from 20 to 336, and 13 categories of FRPN
FM10 1 2 6 3 ranging from 29.85 to 271.5. The cumulative RPN values “RPN-
FM11 8 7 2 5 c”, its percentage “RPN-p”, the FRPN values “FRPN-c”, as well
as its percentage “FRPN-p” are determinate as shown in TABLE
FM12 7 8 9 8
Ⅺ and TABLE Ⅻ.
FM13 5 6 7 7

2021 Global Reliability and Prognostics and Health Management


(PHM-Nanjing)
TABLE X. RISK PRIORITY NUMBER AND ITS RANKING USING Figs.6 and Figs.7 are Pareto chart for RPN and FRPN
TRADITIONAL FMECA AND DOUBLE LAYER FUZZY ENHANCED FMEA.
respectively. It can be observed from Fig.6 that the main causes
Failure
RPN FRPN
of ATO failure are FM2, FM3, FM6, FM7, FM8 and FM13. In
cause S O D RPN FRPN Fig.7, the main causes have 7 main causes, FM3, FM6, FM7,
Ranking Ranking
number FM8, FM9, FM12 and FM13. Obviously, double layer fuzzy
FM1 4.03 5 1 20.15 11 30.78 12 enhanced FMEA can not only find out more failure causes than
FM2 4 5 6 120 5 95.99 8 tradition FMECA but also distinguish which is more important
with the same RPN value.
FM3 7 6 6 252 2 238.5 2
FM4 2 5 2 20 12 29.85 13
FM5 7 2 7 98 6 65.59 9
FM6 8 3 5 120 5 96.14 7
FM7 5 6 7 210 4 161.4 4
FM8 8 7 6 336 1 271.5 1
FM9 6 5.03 3 90.54 7 114.3 5
FM10 3 5 3 45 10 45.16 11
FM11 5 2 5 50 9 61.48 10
FM12 8 5 2 80 8 98.59 6
FM13 7 7 5 245 3 219.9 3

TABLE XI. PARETO ANALYSIS FOR RPN Figure 6. RPN for Pareto chart.

Category RPN RPN-c RPN-p

Category 1 336 336 21.44


Category 2 252 588 37.53
Category 3 245 833 53.17
Category 4 210 1043 66.57
Category 5 120 1163 74.23
Category 6 98 1261 80.49
Category 7 90.54 1351.54 86.27
Category 8 80 1431.54 91.37
Category 9 50 1481.54 94.56
Category 10 45 1526.54 97.44 Figure 7. FRPN for Pareto chart
Category 11 20.15 1546.69 95.72
V. CONCLUSIONS
Category 12 20 1566.69 100
The paper results show that the method based on double
TABLE XII. PARETO ANALYSIS FOR FRPN layer fuzzy enhanced FMEA can improve the accuracy of
Category
finding main failure causes and distinguish which is the more
FRPN FRPN-c FRPN-p important failure cause in the same RPN value. Power source,
Category 1 271.5 271.5 17.75 communication board and input/output unit are frequently faulty
equipment, which should be inspected and repaired immediately.
Category 2 238.5 510 33.35
During finding failure causes, we need pay more attention to the
Category 3 219.9 729.9 47.73 potential causes, which are fault of tachogenerator, traction fault
Category 4 161.4 891.3 58.29 and loss ATP (Automatic Train Protection) signal.
Category 5 114.3 1005.6 65.76
ACKNOWLEDGMENT
Category 6 98.59 1104.19 72.20
This work was supported by the Beijing Municipal Natural
Category 7 96.14 1200.33 78.50
Science Foundation (L191013) and Fundamental Research
Category 8 95.99 1296.32 84.77 Funds for the Central Universities (2020YJS016).
Category 9 65.59 1361.91 89.26
Category 10 61.48 1423.39 93.08
Category 11 45.16 1468.55 96.04
Category 12 30.78 1499.33 98.05
Category 13 29.85 1529.18 100

2021 Global Reliability and Prognostics and Health Management


(PHM-Nanjing)
REFERENCES [10] Power systems reliability subcommittee of the power systems engineering
committee. recommended practice for the design of reliable industrial and
[1] Bin Xie. Research of Fault Diagnosis Expert System for ATO Based on commercial power systems. IEEE Ind Appl Soc IEEE Std 2007.
Fault-Tree[D]. Lanzhou Jiaotong University,2013.
[11] Stamatis, D. H. (1995). Failure mode and effect analysis—FMEA from
[2] Yahmadi Raja and Brik Kais and ben Ammar Faouzi. Fuzzy risk priority theory to execution. New York: ASQC Quality Press.
number assessment for solar gel battery manufacturing defects[J].
Engineering Failure Analysis, 2021, 124(prepublish): 105327-. [12] Mahmood Shafiee and Fateme Dinmohammadi. An FMEA-Based Risk
Assessment Approach for Wind Turbine Systems: A Comparative Study
[3] N.G. Mutlu, S. Altuntas Risk analysis for occupational safety and health of Onshore and Offshore[J]. Energies, 2014, 7(2) : 619-642.
in the textile industry: integration of FMEA, FTA, and BIFPET methods
Int. J. Ind. Ergon., 72 (2019), pp. 222-240 [13] H. Ghasemian, Q. Zeeshan, D.A. Firouzi Failure Mode and Effect
Analysis of Gas Turbine Power Plant Systems (GTPPSs) Eastern
[4] C.-F. Chi, D. Sigmund, M.O. Astardi Classification Scheme for root cause Mediterranean University (2017)
and failure modes and effects analysis (FMEA) of passenger vehicle
[14] Gupta Gajanand and Ghasemian Hamed and Janvekar Ayub Ahmed. A
recalls Reliab. Eng. Syst. Saf., 200 (2020), p. 106929,
novel failure mode effect and criticality analysis (FMECA) using fuzzy
10.1016/j.ress.2020.106929
rule-based method: A case study of industrial centrifugal pump[J].
[5] N. Chanamool, T. Naenna, Fuzzy FMEA application to improve decision- Engineering Failure Analysis, 2021, 123
making process in an emergency department, Appl. Soft Comput. 43
(2016) 441–453. [15] Sharma R. K., Kumar D., Kumar P., (2005), Systematic failure mode
effect analysis (FMEA) using fuzzy linguistic modeling. International
[6] Yu Zang. Remaining useful life prediction and health management of on- Journal of Quality & Reliability Management.
board system equipment in high-speed railways[D]. Beijing Jiaotong
University ,2020. [16] Cicí lia RM Leite et al (2011). A fuzzy model for processing and
monitoring vital signs in ICU patients, BioMedical Engineering OnLine.
[7] Stamatis, D. H. (1995). Failure mode and effect analysis—FMEA from 10:68.
theory to execution. New York: ASQC Quality Press.
[17] J. Balaraju, M.G. Raj, C.S. Murthy Fuzzy-FMEA risk evaluation
[8] C.T. Su, H.C. Lin, P.W. Teng, YT Improving the reliability of electronic approach for LHD machine-A case study J. Sustainable Mining., 18
paper display using FMEA and Taguchi methods: a case study (2019), pp. 257-268
Microelectron. Reliab. (2014), pp. 1369-1377
[18] W. Song, X. Ming, Z. Wu, B. Zhu A rough TOPSIS approach for failure
[9] A.S.T. Maria Jayaprakash Failure detection and optimization of sugar mill mode and effects analysis in uncertain environments Qual. Reliab. Eng.
boiler using FMEA and Taguchi method Eng. Fail Anal. (2013), pp. 17- Int., 30 (4) (2014), pp. 473-486.
26.

2021 Global Reliability and Prognostics and Health Management


(PHM-Nanjing)

You might also like