Professional Documents
Culture Documents
art ic l e i nf o
a b s t r a c t
Article history:
Received 3 June 2013
Received in revised form
23 April 2014
Accepted 27 April 2014
Available online 14 May 2014
The hazard rate function is always applied to make maintenance policy, and the usual hazard rate
function is computed by the data of failure times of systems working in constant environment, thus for
systems working in dynamic environment it cannot be directly applied. In this paper, hazard rate
function of system in the dynamic environment is computed, and the effects of current environment
status and the environmental history on hazard rate function are explicitly presented. For system with
the known degradation process, hazard rate function is studied by the Markov additive process. The
environment evolution process is modeled as a stochastic process with two states, one state represents
normal environment, the other represents severe environment, and system degrades more quickly
under severe environment than under normal environment. The relationship between hazard rate
functions of system in time-invariant and dynamic environment is researched, from which three
important facts are revealed, rstly hazard rate function jumps as the environment jumps, secondly the
form of hazard rate function is determined Wby current environment state, and thirdly the effective age
of system is determined by the environmental history. For system with the unknown degradation
process, based on the above facts, this paper derives the hazard rate function in dynamic environment,
and proposes a method to compute the effective age under given environmental history. Finally the
optimal maintenance policy for system in dynamic environment is studied.
& 2014 Elsevier Ltd. All rights reserved.
Keywords:
Maintenance
Hazard rate function
Dynamic environment
Markov additive process
Effective age
1. Introduction
Maintenance is applied to slow the degradation level of the
system, when making optimal maintenance policy, hazard rate
function (HRF) is always used [5,8,10,12]. Nakagawa [29] dened
the classic HRF mathematically and proposed several classical
maintenance policies based on HRF for non-repairable system.
Also based on the hazard rate function, many imperfect maintenance models for repairable system have also been proposed
[19,24,32]. Lie and Nakagawa extended HRF by introducing the
multiply factor and the effective age adjustment factors [20,27,28].
Furthermore, Lin et al. proposed a hybrid HRF model, in which
both the multiply factor and effective age of the HRF are adjusted
after imperfect preventive maintenance [21,22,34].
However, there is a disadvantage in the above works, that is,
both external environment and operational modes of the system
are time-invariant [11,13,39,40]. External environment means
humidity, temperature, stress, wind speed, radiation of sunlight,
shock and so on; operational mode refers to how the system
Corresponding author.
E-mail address: lium@tsinghua.edu.cn (L. Min).
http://dx.doi.org/10.1016/j.ress.2014.04.020
0951-8320/& 2014 Elsevier Ltd. All rights reserved.
L. XiaoFei, L. Min / Reliability Engineering and System Safety 130 (2014) 5060
Nomenclature
HRF
MAP
Notations
Xt
X threshold
Tf
z1
z2
r 1 t
r 2 t
51
zi ; t 1 -zj ; t 2 - environment evolution process, the environment is zi during 0; t 1 , then changes into zj during
t 1 ; t 1 t 2 , and so on
P
environmental process
H
history of external environment
h1 tjH HRF of system at t, the system is in dynamic environment with history H, and environment is normal at t
h2 tjH HRF of system at t, the system is in dynamic environment with history H, and environment is severe at t
1
parameter of Poisson process of system in normal
environment
2
parameter of Poisson process of system in severe
environment
52
L. XiaoFei, L. Min / Reliability Engineering and System Safety 130 (2014) 5060
X 0 1; 1 t
X0!
h1 (t | H 2 )
r1 (t )
h1 (t | H 3 )
t1
t1 + t2
h1 (t | H 4 )
t2
t3
t4
t1
t
Fig. 6. Environmental process of Scene D.
L. XiaoFei, L. Min / Reliability Engineering and System Safety 130 (2014) 5060
53
h1 (t | H 5 )
t2
t3
t4
t5
t1
t
Fig. 7. Environmental process of Scene E.
R1 0
Here X threshold X 0 1, X 0 1; 1 t 1 t uX e u du, and
d
X 0 1; 1 t 1 t
d t
Z 1
0
0
d
uX e u du 1 1 t 1 tX e 1 t 1 t
dt 1 t 1 t
r 1 t lim
lim
r 1 t lim
PX t r X jH 1 PX t t r X jH 1 -z1 ; t
PX t rX 0 jH 1 t
0
t-0
2d=dtPX t t r X 0 jH 1 -z1 ; t
PX t r X 0 jH 1
t-0
p
2d=dtX 0 =s1 t t
p
lim
t-0
2X 0 =s1 t 1
p 0
p
02
2
1= 2 X =s1 t 3=2 e X =2s1 t =2X 0 =s1 t 1
t-0
X 0 1; 1 t X 0 1; 1 t 1 t
X 0 1; 1 tt
t-0
d=dt X 0 1; 1 t 1 t
lim
X 0 1; 1 t
t-0
0
1 1 tX e 1 t
lim
lim
R1
X u
1 t u e du
PT f A t; t tjH 2
0
lim
R1
X u
2 t u e du
t-0 PX t r X jz2 ; tt
X 0 1; 2 t X 0 1; 2 t 1 t
lim
X 0 1; 2 tt
t-0
0
d
X 1; 2 t 1 t
lim dt
X 0 1; 2 t
t-0
X 0 2 t
1 2 t e
h1 tjH 2 lim
PX t r X 0 jH 1 PX t t r X 0 jH 1 -z1 ; t
PX t r X 0 jH 1 t
Let 1 t 1 2 t 2 , then
PT f A t; t tjH 3
PX t r X 0 jH 3 t
h1 tjH 3 lim
t-0
X 0 1; X 0 1; 1 t
X 0 1; t
t-0
lim
R 1 t X 0 u
u e du
lim R 1 X 0 u
t-0 t u e du
0
1 1 t 1 2 t 2 X e 1 t1 2 t2
R1
X u
1 t 1 2 t 2 u e du
r 1 t 1 t 2
1 1 t 1 1 t 2 X e 1 t1 1 t2
R1
X u
1 t 1 1 t 2 u e du
10
54
L. XiaoFei, L. Min / Reliability Engineering and System Safety 130 (2014) 5060
h1 tjH 5 lim
t-0
1 1 t 1 2 t 2 1 t 3 2 t 4 1 t 5 X e 1 t1 2 t2 1 t3 2 t4 1 t5
R1
X u
1 t 1 2 t 2 1 t 3 2 t 4 1 t 5 u e du
13
By Eqs. (4), (9), (12) and (13), HRF of system in dynamic
environment can be summarized as for Zt z1
0
1 X e
h1 tjH R 1
14
X u
u e du
2 X e
h2 tjH R 1
15
X u
u e du
0
0
0
r 1 t
t 1 t 2 X
1 X e 1 t2 = U t t uX e u du
2
mi s t mi sn
exp mi s t mi s
n!
16
11
PX t o X threshold
X 0 1; mi t
17
X0!
R1
1 t tmi t tX e mi t t
Let i ij 1 t j , then
r 1 t
Let 1 t 1 2 t 2 1 t 3 2 t 4 ,
1 tm1 tX e m1 t
R1
18
m1 t
uX e u du
0
PT f A t; t tjH 4
0
t-0 PX t rX jH 4 t
h1 tjH 2
h1 tjH 4 lim
1 tm2 tX e m2 t
R1
m2 t
R 1 t X 0 u
u e du
lim R 1 X 0 u
t-0 t u e du
19
0
uX e u du
h1 tjH 3
1 m1 1 m2 1 ; 2 X e m1 1 m2 1 ;2
R1
m1 1 m2 1 ;2
uX e u du
20
1 1 t 1 2 t 2 1 t 3 2 t 4 X e 1 t1 2 t2 1 t3 2 t4
R1
X u
1 t 1 2 t 2 1 t 3 2 t 4 u e du
12
It is obvious that Eqs. (18)(22) are similar to Eqs. (3), (4), (9),
(12) and (13). HRF of system in dynamic environment can be
L. XiaoFei, L. Min / Reliability Engineering and System Safety 130 (2014) 5060
55
h1 tjH 4
1 m1 1 m2 1 ; 2 m1 2 ; 3 m2 3 ; 4 X e m1 1 m2 1 ;2 m1 2 ;3 m2 3 ;4
R1
21
m1 1 m2 1 ; 2 m1 2 ;3 m2 3 ; 4
uX e u du
0
h1 tjH 5
1 m1 1 m2 1 ; 2 m1 2 ; 3 m2 3 ; 4 m1 4 ; 5 X e m1 1 m2 1 ;2 m1 2 ;3 m2 3 ;4 m1 4 ;5
R1
m1 1 m2 1 ;2 m1 2 ; 3 m2 3 ;4 m1 4 ; 5
summarized as for Zt zi
X e
23
X u
u e du
Rt
1 1 tX e 1 t
r 1 t R 1 X 0 u
1 t u e du
0
1 X e
h1 tjH R 1
X u
u e du
Since
1 1 2 2 1 1 2 2
1
24
1 X e
h1 tjH R 1
0
uX e u du
1 1 A1 X e 1 A1
R 1 X0 u
r 1 A1
1 A1 u e du
25
2 X e
2 2 A2 X e 2 A2
R 1 X0 u
r 2 A2
X e u du
u
2 A2 u e du
h2 tjH R 1
22
hi tjH R 1i
uX e u du
26
56
L. XiaoFei, L. Min / Reliability Engineering and System Safety 130 (2014) 5060
t-0
N s t; t t
27
t-0M N s 0; tt
28
h1 tjH
1 1 X e 1
R1
X u
1 u e du
1 e 1
R1
u
1 e du
1 e 1
e 1
1 :
h2 tjH
2 2 X e 2
R1
X u
2 u e du
2 e 2
R1
u
2 e du
2 e 2
e 2
N d t; t t
t-0M N d 0; tt
lim
N z1 t; t t
t-0M N d 0; tt
Nz1 t; t t
t-0M N d 0; tt
lim
N z1 t; t t
t-0M N z1 0; A1 t
30
31
For H z1 ; t 1 z2 ; t n -z1 ; , let Y H g 1 z1 ; 1 g 2 z1 ; 2 ,
then the hazard rate at t t 1 t 2 is
Rd tjH Rd t tjH-z1 ; t
Rd tjHt
Y H Y H g 1 z1 ; t
lim
Y H t
t-0
du
dg 1 z1 ; u
1
du
du
Y
h1 tjH lim
t-0
29
u YH
u0
L. XiaoFei, L. Min / Reliability Engineering and System Safety 130 (2014) 5060
d ln u
dg 1 z1 ; u
du u Y H
du
u0
p
2Y H 2
p p
p
p
p
2 2 1 t 1 2 t 2 1 t n 2
0s
1
s
32
u YH
u 0 Y
du
du
H
d ln u
dg 2 ; u
u Y H 2
du
du
57
1
1
t t
t A
2 1 2
2 n
2 2 @
39
h2 tjH lim
t-0
40
s
X2
33
u0
35
1
1
t t
t
2 1 2
2 n
By Eqs. (37) and (40), the HRF of system in dynamic environment has the similar form as Fig. 11 shows.
Example 3. r 1 t 61 t t 2 ; r 2 t 62 t t 2 .
The survival functions of system at normal and severe environment are
p 2
p
3
(
3
2
3
R1 t e 31 t 21 t e 3 1 t 3 1 t
p 2
p
3 :
42
3
2
3
R2 t e 32 t 22 t e 2 2 t 2 2 t
p
p
2
3
By Eq. (40), u; v e 3u e 2v , 1 t and 2 t can be con 3u2
sidered
asp
the
;
p
contribution of section of z1 ; t and z2 ; t to e
3
1 t and 3 2 t can be considered as the contribution of section of
3
z1 ; t and z2 ; t to e 2v , thus the contribution of zi ; t with
arbitrary initial time to the survival function of system is the
same, and the contribution of dynamic environment is divided
into two parts, Y 1H and Y 2H .
For H z1 ; t 1 z2 ; t n -z1 ; ,
(
p
p
p
p
Y 1H 1 t 1 2 t 2 2 t n 1
43
p
p
p
p
Y 2H 3 1 t 1 3 2 t 2 3 2 t n 3 1 :
The survival function at t i t i is
2
RtjH e 3Y 1H e 2Y 2H
The HRF at t t 1 t 2 is
h1 tjH lim
Rd tjH Rd t tjH-z1 ; t
Rd tjHt
p 2
2
e Y H e Y H 1 t
lim
lim
lim
e 3Y 1H e 2Y 2H
t-0
p
p
3
6 1 Y 1H 6 1 Y 2H 2
0
1
s
2
2
t
t A
21 @t 1
1 2
1 n
36
61 @t 1
2
3 2
t A 6 1 @ t 1
t
1 2
1 2
!2
45
37
PX t o X 0 jH PX t t o X 0 jH-z2 ; t
PX t o X 0 jHt
p 2
p
3
3
3Y 1H 2 2Y 2H 3
e
e
e 3Y 1H 2 t e 2Y 2H 2 t
h2 tjH lim
t-0
2
2
t
t
X1 t1
1 2
1 n
For
H 1 ; t 1 1 ; t n -2 ; ,
t 1 t 2 t is
e 3Y 1H e 2Y 2H t
p 2
p
3
p
p
p
p
3
1 Y 1H 1 t 3 1 Y 2H 3 1 t2 e 3Y 1H 1 t e 2Y 2H 1 t
2
t-0
2
e YH
p
2Y H 1
p p
p
p
p
2 1 1 t 1 2 t 2 2 t n 1
0
1
s
s
PX t o X 0 jH PX t t o X 0 jH-z1 ; t
PX t o X 0 jHt
p 2
p
3
3
2
3
e 3Y 1H e 2Y 2H e 3Y 1H 1 t e 2Y 2H 1 t
t-0
t-0
44
h1 tjH lim
e Y H t
p 2
p
p
2Y H 1 t 1 e Y H 1 t
t-0
The HRF at t t 1 t 2 is
t-0
lim
41
38
lim
lim
the
R tjH Rd t tjH-z2 ; t
h2 tjH lim d
Rd tjHt
t-0
p
Y 2H
Y H 2 t2
e
e
lim
2
t-0
e Y H t
hazard
rate
at
e 3Y 1H e 2Y 2H t
p 2
p
3
p
p
p
3
2 Y 1H 2 t 3 2 Y 2H 3 2 t2 e 3Y 1H 2 t e 2Y 2H 2 t
2
t-0
e 3Y 1H e 2Y 2H
t-0
p
p
3
6 2 Y 1H 6 2 Y 2H 2
0s
1
62 @
0s
1
3 1
t t A 6 2 @
t t
2 1 2
2 1 2
!2
46
58
L. XiaoFei, L. Min / Reliability Engineering and System Safety 130 (2014) 5060
By Eqns. (43) and (44), the HRF of system in dynamic environment has the similar form as Fig. 11 shows.
The HRF of system in constant environment and the one in
dynamic environment can be integrated into one function r t; ; z,
where t is the operation time, is the total time of severe
environment, z is the condition of environment at t. Then
8
rt; 0; 0 r 1 t
>
>
>
>
<
rt; 0; 1 r 2 t
47
;
0
h
rt;
>
1 tjH; H z2 ; -z1 ; t
>
>
>
: rt; ; 1 h2 tjH; H z1 ; t -z2 ;
Remark 1. When the contributions of zi ; t with different initial
times to survival function are not the same, according to the
denition of HRF Eq. (27), the HRF of system in dynamic environment also jumps between r 1 t and r 2 t when the external
environment changes, except it is difcult to determine the virtual
age of system.
process is Q
, where 1=; 1=are expected
sojourn time in normal and severe state respectively. Then,
RT
s-expected cost during 0; T is cr cm 0 htdt, where ht is the
hazard rate function of system in dynamic environment at time t.
Then, the s-expected cost rate is
RT
cr cm 0 htdt
48
T
d cr cm 0 htdt
dT
T
cm hTT cr cm
T2
RT
0
49
1
1
r2
x t x f 12 t; xdx
p1
2
0
Z t
2
p2
r 1 x t x f 21 t; xdx
1
0
Z
51
52
2k r
z=2
Here I r z 1
k 0 k!k r! is the modied Bessel function of order r
and
p p
p
f 1 t; x ft x I 0 2 xt x x=t xI 1 2 xt xge x t x
p p
p
f 2 t; x fx I 0 2 xt x t x=xI1 2 xt xge x t x
53
q
f 21 t; x I 0 2 xt xe x t x
54
q
f 12 t; x I 0 2 xt xe x t x
55
f 22 t; x xe t
ht
q
q
t x=xI 1 2 xt xe x t x
56
t
t
e
r 1 t
e
r 2 t
r2
htdt
1
x t x f 22 t; xdx
2
0
Z t
2
r 1 x t x p1 f 11 t; x p2 f 21 t; xdx
1
0
Z t
1
r2
x t x p1 f 12 t; x p2 f 22 t; xdx
2
0
By analysis, we found
q
q
f 11 t; x t xe t x=t xI 1 2 xt xe x t x
RT
p2
r 1 @x
0s
q
2
t xA
I 0 2 xt xe x t x dx
1
q
1
x t xA
r2 @
I 0 2 xt xe x t x dx
2
1 p
2
x=t x
t xA
r 1 @x
I1
1
q
2 xt xe x t x dx
0s
1 p
Z t
1
t x=x
@
x t xA
r2
2
0
q
I 1 2 xt xe x t x dx
57
For simplicity,
Let
p
p
1, also substitute r 1 t t, r 2 t 4t,
2 =1 2, 1 =2 1=2 into ht, we obtain
1
1
ht e t t e t 4t
2
2
Z t
p
1
x 2t x I 0 2 xt xe t dx
2
0
Z t
p
1
1
4 x t x I 0 2 xt xe t dx
2
2
0
p
Z t
p
x=t x
I 1 2 xt xe t dx
x 2t x
2
0
p
Z t
p
t x=x
1
I 1 2 xt xe t dx
4 x t x
2
2
0
L. XiaoFei, L. Min / Reliability Engineering and System Safety 130 (2014) 5060
59
Z t
p
et
5t
6t 3xI 0 2 xt xdx
2
0
Z
2t x
p
p
p
x=t x 4t 2x t x=xI 1 2 xt xdx
58
With Eq. (58), the s-expected cost versus T is computed and
plotted in Fig. 15, the optimal preventive interval is T 2:1, which
corresponds to minimal s-expected cost rate. We also computed
the s-expected cost rate by Monte Carlo Simulation (MCS), Fig. 16
shows the ow chart, and the results is also plotted in Fig. 15. The
s-expected cost rate computed by Eq. (48) and Monte Carlo
Simulation are the same.
Fig. 16. Flow chart of computing s-expected cost rate of system in dynamic
environment.
Acknowledgments
7. Conclusions
In this paper, the HRF of system in dynamic environment is
computed. By analyzing the relationship of HRF of system in timeinvariant and the one in dynamic environment, three facts are
found out, rstly the HRF of system jumps when the external
environment changes, secondly the HRF form is determined by
current environment condition, and thirdly the effective age is
determined by environmental history. For system which has
constant HRF in time-invariant environment, the HRF is computed
by considering degradation process as a special Markov additive
process. For system with unknown degradation process, HRF of
system in dynamic environment has the same property with HRF
researched by Markov additive process, but the effective age is
difcult to be computed when external environment changes.
Section 4 proposes the conditions under which the effective age
can be analytically computed. Finally maintenance policy for
system in dynamic environment is optimized based on HRF
studied in this paper.
References
[1] Almalki SJ, Nadarajah S. Modications of the Weibull distribution: a review.
Reliab Eng Syst Saf 2014;124:3255.
[2] Barker K, Baroud H. Proportional hazards models of infrastructure system
recovery. Reliab Eng Syst Saf 2014;124:2016.
[3] Bjerga T, Aven T, Zio E. An illustration of the use of an approach for treating
model uncertainties in risk assessment. Reliab Eng Syst Saf 2014;125:4653.
[4] Bris R, Medonos S, Wilkins C, Zdrhala A. Time-dependent risk modeling of
accidental events and responses in process industries. Reliab Eng Syst Saf
2014;125:5466.
[5] Cha JH, Finkelstein M. The failure rate dynamics in heterogeneous populations.
Reliab Eng Syst Saf 2013;112:1208.
[6] inlar E. Markov additiveprocesses II. Z Wahrscheinlichkeitstheorie Verw Geb
1972;24:95121.
60
L. XiaoFei, L. Min / Reliability Engineering and System Safety 130 (2014) 5060
[7] Cox DR. Regression models and life tables. Journal of the Royal Statistical
Society. B. 1972;34:187202.
[8] Cui L, Kuo W, Loh HT, Xie M. Optimal allocation of minimal and perfect repairs
under resource constraints. IEEE Trans Reliab 2004;53:1939.
[9] Delia MC, Rafael PO. Matrix stochastic analysis of the maintainability of a
machine under shocks. Reliab Eng Syst Saf 2014;121:117.
[10] Doyen L, Gaudoin O. Classes of imperfect repair models based on reduction of
failure intensity or virtual age. Reliab Eng Syst Saf 2004;84:4556.
[11] Elferik S, Ben-Daya M. Age-based hybrid model for imperfect preventive
maintenance. IIE Trans 2006;38:36575.
[12] Finkelstein M. Failure rate modeling for reliability and risk. London: SpingerVerlag; 2008.
[13] Garca-Daz JC, Gozalvez-Zafrilla JM. Uncertainty and sensitive analysis of
environmental model for risk assessments: an industrial case study. Reliab
Eng Syst Saf 2012;107:1622.
[14] Ghasemi A, Yacout S, Ouali MS. Evaluating the reliability function and the
mean residual life for equipment with unobservable states. IEEE Trans Reliab
2010;59:4554.
[15] Jardine AKS, Anderson PM, Mann DS. Application of the weibull proportional
hazards model to aircraft and marine engine failure data. Qual Reliab Eng Int
1987;3:7782.
[16] Jardine AKS, Banjevic D, Wiseman M, Buck S, Joseph T. Optimizing a mine haul
truck wheel motors' condition monitoring program. J Qual Maint Eng
2001;7:286301.
[17] Jiang XM, Yuan Y, Liu X. Bayesian inference method for stochastic damage
accumulation modeling. Reliab Eng Syst Saf 2013;111:12638.
[18] Kumar D, Westberg U. Proportional hazards modeling of time-dependent covariates using linear regression: a case study. IEEE Trans Reliab 1996;45:38692.
[19] Levitin G, Lisnianski A. Optimization of imperfect preventive maintenance for
multi-state systems. Reliab Eng Syst Saf 2000;67:193203.
[20] Lie CH, Chun YH. An algorithm for preventive maintenance policy. IEEE Trans
Reliab 1986;35:715.
[21] Lin D, Zuo MJ, Yam RCM. General sequential imperfect preventive maintenance models. Int J Reliab Qual Saf Eng 2000;7:25366.
[22] Lin D, Zuo MJ, Yam RCM. Sequential imperfect preventive maintenance models
with two categories of failure modes. Naval Res Logist 2001;48:17283.
[23] Lu XF, Chen MY, Liu M, Zhou DH. Optimal imperfect periodic preventive
maintenance for systems in time-varying environments. IEEE Trans Reliab
2012;61:42639.
[24] Lugtigheid D, Jiang X, Jardine AKS. A nite horizon model for repairable
systems with repair restrictions. J Oper Res Soc 2008;59:132131.
[25] Mattrand C, Bourinet JM. The cross-entropy method for reliability assessment
of cracked structures subjected to random Markovian loads. Reliab Eng Syst
Saf 2014;123:17182.
[26] McKinlay SA. Markov-modulated models for derivatives pricing [Postgraduate
Diploma Thesis]. Australia: The University of Melbourne; 2009.
[27] Nakagawa T. A summary of imperfect preventive maintenance policies with
minimal repair. RAIRO Oper Res 1980;14:24955.
[28] Nakagawa T. Sequential imperfect preventive maintenance policies. IEEE Trans
Reliab 1988;37:2958.
[29] Nakagawa T. Maintenance theory of reliability. Springer-Verlag London
Limited; 2005.
[30] Park JH, Chang W, Lie CH. Stress-reducing preventive maintenance model for a
unit under stressful environment. Reliab Eng Syst Saf 2012;108:428.
[31] Peysson F, Ouladsine M, Outbib R. A generic prognostic methodology using
damage trajectory models. IEEE Trans Reliab 2009;58:27785.
[32] Pham H, Wang H. Imperfect maintenance. Eur J Oper Res 1996;94:42538.
[33] Samrout M, Chatelet E, Kouta R, Chebbo N. Optimization of maintenance
policy using the proportional hazard model. Reliab Eng Syst Saf
2009;94:4452.
[34] Sheu SH, Chang CC. A periodic replacement model based on cumulative
repair-cost limit for a system subjected to shocks. IEEE Trans Reliab
2010;59:37482.
[35] Singpurwalla ND. Survival in dynamic environments. Stat Sci 1995;10:86103.
[36] Vlok PJ, Goetzee JL, Jardine AKS, Makis V. Optimal component replacement
decisions using vibration monitoring and the proportional hazards model.
J Oper Res Soc 2002;53:193202.
[37] Wang XL, Balakrishnan N, Guo B. Residual life estimation based on a generalized Wiener degradation process. Reliab Eng Syst Saf 2014;124:1323.
[38] Wang ZQ, Wang PF. A new approach for reliability analysis with time-variant
performance characteristics. Reliab Eng Syst Saf 2013;115:7081.
[39] Wu S, Clements-Croome D. Optimal maintenance policies under different
operational schedules. IEEE Trans Reliab 2005;54:33846.
[40] Wu SM, Zuo MJ. Linear and nonlinear preventive maintenance models. IEEE
Trans Reliab 2010;59:2429.