You are on page 1of 4

Mission reliability assessment for plane fleet based on

flight profile

Liu Zhenyu, Ma Xiaobing, Zhao Yu
School of Reliability and Systems Engineering
Beihang University
Beijing, China
liuzhenyu@dse.buaa.edu.cn


AbstractCombat mission of battle plane is usually executed by
plane fleet as a unit. Plane fleet can be divided into several team
formations. Battle planes in different formations have different
flight profiles, and cooperate with each other. The load
conditions varied under different flight profiles. According to the
influence of load conditions on battle planes reliability, the flight
profile conversion factor was introduced to establish the
reliability model of battle plane. Then, the fault data were
utilized synthetically to measure fault rate under each flight
profile. On the basis, the mission reliability model of plane fleet
was derived. Based on the model, the statistical inference method
of mission reliability for plane fleet was put forward using the
fault rate under each profile. The illustrative example shows that
the method is reasonable and straightforward for engineering.
Keywords-battle plane; flight profile; reliability assessment;
profile conversion factor;
I. INTRODUCTION
Combat missions are usually executed by many planes as a
fleet. Planes in the fleet are divided into several teams. Each
team has its own subtask and cooperates with each other. Each
subtask corresponds to a flight profile. [1] The load conditions
varied under different flight profiles. Therefore, the difference
among flight profiles should be considered in mission
reliability assessment.
Considering the features of battle plane, its fault data is
collected after a given period of time. During this period, battle
plane may execute many training missions. The assessment
method currently cannot take into account the influence that
caused by different load conditions, only appear reliability
average. The reason is inadequate use of hidden information in
fault data. In fact, the influence of cumulative flight time under
each flight profiles on fault is independent. Conventional
method does not differentiate among flight time under each
flight profiles. Thus reliability model of fault data is not
accurate enough and some fault information is lost.
This paper proposed a new approach to assess mission
reliability of plane fleet. That is, analyze it in the view of flight
profile. Then, reliability model is established after studying the
influence of load conditions on battle plane. The fault rates
were measured by using fault data synthetically under each
flight profiles. On the basis, the mission reliability model of
plane fleet was derived. Based on the model, the statistical
inference method of mission reliability for plane fleet was put
forward.
II. FLIGHT PROFILE CONVERSION FACTOR
A. Define of Flight Profile
Flight profile is a graphical step-by-step depiction of
trajectory or phase of flight, followed by altitude, speed, and
flight time [2]. According to mission requirements, battle
plane's flight profiles mainly include interception, air
superiority, ground attack and reconnaissance. Flight profile
consists of several mission sections, including takeoff,
climbing, cruising, air-to-air, air-to-land, landing, etc. As
shown in Fig. 1, commonly used flight profile includes all or
part of the mission sections.
H,km
10 20 30 40
2
4
6
8
0
cruising
climbing
T,min
landing

Figure 1. Typical flight profile
B. Define of Flight Profile Conversion Factor
The load conditions varied under different flight profiles, in
which altitude, velocity and acceleration are very different.
These three parameters determine the pressure, vibration and
overload, which affect the reliability of battle plane. The air
pressure environment of equipment in battle plane mainly
depends on altitude. In the air, the pressure is lower than stand
atmospheric pressure, low air pressure may cause
physical/chemical effects, thermal effects and electrical effects.
These can result in airborne equipment failure, thus affect
mission reliability. Vibration is another important factor that
affects battle plane's reliability. Vibration is mainly caused by
dynamic pressure, while the dynamic pressure is determined by
flight altitude and velocity [3]. Overload which is caused by
acceleration has great effect on many airborne equipment
including machinery, electronics and materials.
Identify applicable sponsor/s here. If no sponsors, delete this text box.
(sponsors)
Therefore, battle plane's reliability varied under different
flight profiles. In order to comprehensive utilize fault data
under different flight profiles; we defined flight profile
conversion factor referring to the concept of environment
conversion factor [4, 5]. Assume that reliability functions are
( )
1
R t and ( )
2
R t respectively in environment I and II, both
have the same distribution type. The conversion factor k ,
when follows the criteria ( ) ( )
1 1 2 2
R t R t = , is
( ) ( )
2
1
1 1 2 2
. :
t
k
t
st R t R t
=
=
(1)
That is to say, life
1
t under environment I equals life
1
kt under
environment II.
Battle plane can be regarded as a complex maintainable
system, assuming that intervals of time between faults are all
exponential distributions under different flight profiles. Choose
the most commonly used flight profile as baseline profile
0
x ,
where baseline fault rate is
0
and reliability function is given
by
0 0
0 0
( )
t
R t e

= . Similarly, under flight profile
i
x , fault rate
is
i
and reliability function is given by ( )
i i
t
i i
R t e

= . Based
on above-mentioned assumption, the conversion factor
i
k
between
i
x and
0
x is obtained by substituting
0 0
( ) R t and
( )
i i
R t into Eq. (1).
0
i
i
k

= (2)
III. MODELING OF MISSION RELIABILITY FOR PLANE FLEET
A. Reliability model of fault data
There are many flight profiles, but some of them are similar.
Profiles can be classified according to the following principles
1) Missions are similar under the same type of profile.
2) Under the same type of profile, flight altitude, velocity
and overload have similar trends.
3) Under the same type of profile, flight time in one
mission approximate to that of other missions.
According to the principles above, flight profiles can be
divided into n categories
T
1 2
( , , , )
n
x x x = X .
i
x denote the
i th profile. Different from conventional fault data, battle
planes fault data is collected after a given period of time.
During this period, battle plane may execute many training
missions. We do not know the fault occurred in which mission.
Suppose within a period of
s
t , m planes executed many
training missions. Denote the fault data as follows
( )
j j
f T , , X ( 1, 2, , ) j m = (3)
where
T
1 2
( , , , )
j j j jn
t t t = T is flight time of the j th plane
under profile
T
1 2
( , , , )
n
x x x = X .
j
f is the number of the j th
planes fault.
Convert flight time under other profiles into baseline profile
0
x . This yield the reliability model of fault data as follows
0
1
n
i ji j j
i
k t f c
=
= +

( 1, 2, , ) j m = (4)
where
j
c is zero mean variable,
0
is unknown. The
conversion factor
i
k in Eq. (2) can be substitute into Eq. (4).
The model can then be written as
1
n
i ji j j
i
t f c
=
= +

( 1, 2, , ) j m = (5)
Let
T
1 2
( , , , )
m
= B T T T ,
T
1 2
( , , , )
m
f f f = F ,
T
1 2
( , , , )
n
= ,
T
1 2
( , , , )
m
c c c = . Eq. (5) may be
reduced to the much more condensed matrix form.
= + B F (6)
B. Mission reliability model of plane fleet
Suppose a plane fleet consists of N planes and can be
divided into
s
n team formations. Planes in the same team
execute same mission. That is, they have same flight profile.
Fig. 2 showed a typical mission. A five planes fleet is divided
into three teams, and bomb three targets respectively.

Figure 2. Diagram of typical mission
Let
1
s
n
i
i
N n
=
=

, where
i
n denote the total number of plane in
the i th profile.
i
m is the mission required minimum number of
plane, and
i i
m n s . Mission success or not is depend on the
number of plane that finished mission. If the number is equal to
or greater than
i
m , mission success. The reliability equation of
plane fleet can be derived as
( )
1
( ) 1 ( )
s i
i
i
i
n n
n k
k k
n i i i i
k m i
R C R t R t

= =
=
I
(7)
here, ( )
i i
R t is the reliability of plane under flight profile
i
x ,
i
t is mission time of flight profile
i
x .
IV. MISSION RELIABILITY ASSESSMENT FOR PLANE FLEET
According to Gauss-Markov theorem, we get the
expectation of fault rate
T 1 T

( )

= B B B F (8)
We assume that is error with (0,1) N , then

follows
( )
2 T 1
, ( ) N o

B B , where
2
2 2

( ) m n
o

=

F B
is unbiased
estimation of
2
o .
2
2
2
1 1

m n
i ji j
j i
t f
= =
| |
=
|
\ .

F B .

Set mission time
i
t , then ( ) exp( )
i i i i
R t t = can be
substituted into Eq. (7). The reliability equation can be derived
as
( )
1
exp( ) 1 exp( )
s i
i
i
i
n n
n k
k
n i i i t
k m i
R C k t t

= =
=
I
(9)
In engineering applications, we must ordinarily obtain
reliability confidence lower limit. Because the distribution of
reliability is hard to get, traditional interval estimate method
cannot be directly deal with. Moreover, approximate interval
estimation is easier to obtain. In most cases, there is no
significant difference between them. So we use approximate
interval estimation instead of accurate interval estimation [6].
For mission reliability model of plane fleet shown in Eq. (9),
asymptotic variance of mission reliability can be expressed as
2
2
1
var( )
s
n
R j
j j
R
o

=
| |
c
= |
|
c
\ .

(10)
Let
2

o denote the variance of

j
, the estimator of
2
R
o
can be obtained by
2
2 2

1
|
s
j
j j
n
R
j j
R


o o

=
=
| |
c
= |
|
c
\ .

(11)
Assume that the estimator of fleets mission reliability
follows asymptotic normal distribution
2
( , )
R
N R o . For a fixed
confidence level 1 o , reliability confidence lower limit of
plane fleet is

L R
R R z
o
o = (12)
where z
o
is the o quantile of standard normal distribution.
V. EXAMPLE
According to the principles of classification in Section III,
flight profiles of a certain type of battle plane can be divided
into four categories
T
1 2 3 4
( , , , ) x x x x = X .
T
z1 z2 z3 z4
( , , , ) t t t t = T
denote flight time under each profile, f denote number of fault.
Fault data of totally six battle planes were collected and
presented in Table I
TABLE I. FAULT DATA OF BATTLE PLANE
No tz1/h tz2/h tz3/h tz4/h f
1
3.6 12.8 7.77 21.26 15
2
10.15 14.13 29.83 1.46 13
3
18.76 33.42 24.66 9.89 22
4
12.86 67.91 54.78 0 27
5
3.71 10.53 12.82 13.14 12
No tz1/h tz2/h tz3/h tz4/h f
6
7.93 12.4 10.43 12.99 12
According to mission scenario, a fleet mission contains
three subtasks corresponding to flight profiles
T
1 2 3
( , , ) x x x .
Fleet is divided into three teams
T
1 2 3
( , , ) f f f , who carry out
their own flight profiles. The required minimum numbers of
battle plane to finish subtask
T
1 2 3
( , , ) x x x are
T T
1 2 3
( , , ) (2,1, 3) m m m = , total numbers of battle plane are
T T
1 2 3
( , , ) (4, 2, 4) n n n = .
(1) Reliability assessment for single battle plane
After substitute ( )
T
15,13, 22, 27,12,12 = F and
3.6 21.26
7.93 12.99
| |
|
=
|
|
\ .
B into Eq. (8) with the fault data in
TABLE I, it is convenient to get estimation of fault rates
( )
T

0.3456, 0.1559,0.2202, 0.4595 = and estimation of


variances ( )
T
0.0680, 0.0347,0.0436, 0.0312 =

.
(2) Reliability assessment for plane fleet
Substitute ( )
T

0.3456, 0.1559,0.2202 = into Eq. (9), we


get the reliability estimation 0.8946 R = with
T T
1 2 3
( , , ) (0.6, 0.6, 0.6) t t t = . Moreover, with

and

, we
get 0.0518
R
o = from Eq. (11). For a fixed typical level
0.1 o = , the percentile of normal distribution is 1.2816 z
o
= .
Finally, we get 0.8282
L
R = using Eq. (12).
VI. CONCLUSION
According to the relation of flight profiles and mission
reliability of battle-plane, the flight profile conversion factor is
introduced to establish the reliability model of fault data. Based
on the model, the statistical inference method of fault rate for
battle plane is proposed. The prediction method of reliability
under the scenario mission is put forward using the fault rates
under typical profiles. This method can consider the influence
of load conditions under different flight profiles. Thus, the
precision of reliability assessment was improved with the
method.
REFERENCES
[1] Ai Jianliang, Qian Guohong, A method to analyze operational
efficiency of the grouped attackers, Fire Control & Command Control,
vol. 25, no. 3, pp. 9-12, 2000
[2] Song Yingdong, Gao Deping, The principal component analysis
method for engine flight mission profiles categorization, Journal of
Aerospace Power, vol. 17, no. 2, pp. 196-200, 2002.
[3] Hong Dongpao, Ma Xiaobing, Zhao Yu and Zhang Lin, Environment
referring factor based on proportional hazards model, Journal of
Beijing University of Aeronautics and Astronautics, vol. 36, no. 4, pp.
443-446, 2010.
[4] Elsayed E. A, Wang H. Z, Bayes & Classical Estimation of
Environmental Factors for the Binomial Distribution, IEEE
Transactions on Reliability, vol. 4, pp. 661-665, 1996.
[5] Sohn S Y, Yoon K B, Chang I S, Random effects model for the
reliability management of modules of a fighter aircraft,. Reliability
Engineering & System Safety, vol. 91, no. 4, pp. 433-437, 2006.
[6] Easterling R G, Approximate confidence limits for system reliability,
Journal of the American Statistical Association, vol. 47, pp. 220-222,
1972.

You might also like