Professional Documents
Culture Documents
Research Paper
Research Paper
Abstract
With control using redundant multiple control surface arrangement and large-deflection drag rudders, a combat flying wing
has a higher probability for control surface failures. Therefore, its flight control system must be able to reconfigure after such
failures. Considering three types of typical control surface failures (lock-in-place (LIP), loss-of-effectiveness (LOE) and float),
flight control reconfiguration characteristic and capability of such aircraft types are analyzed. Because of the control surface
redundancy, the aircraft using the dynamic inversion flight control law already has a control allocation block. In this paper, its
flight control configuration during the above failures is achieved by modifying this block. It is shown that such a reconfigurable
flight control design is valid, through numerical simulations of flight attitude control task. Results indicate that, in the circum-
stances of control surface failures with limited degree and the degradation of the flying quality level, a combat flying wing
adopting this flight control reconfiguration approach based on control allocation could guarantee its flight safety and perform
some flight combat missions.
Keywords: flight control; reconfiguration; control allocation; control surface failure; flying wing; multiple control surfaces; drag
rudder
control allocation. This paper is aimed at studying ent functionality constraints, and then the reconfigura-
principle, implement and validity of this approach. The tion is achieved by certain fault-tolerant algorithms.
results can be used for references in the design re- Generally speaking, in the case of the conventional
searches of the reconfigurable flight control system for configuration, roll control is fulfilled by its aileron,
such novel configuration aircraft types. pitch control by elevator, and yaw control by rudder.
After the aileron fails, the differential deflection of the
2. Aircraft Description elevator can achieve reconfiguration. When the eleva-
tor is in a failure situation, the symmetric deflection of
The typical combat flying wing is designed for the the aileron can provide a pitching moment, so as to
tactical attack mission, with the ability to perform mis- guarantee certain pitch control capability. However,
sions of air-to-air combat, ground attack, etc. Com- when the rudder fails, the aircraft can fulfill yaw con-
pared with the conventional aircraft, it is equipped with trol only through rolling to generate a side force.
more control surfaces (see Fig. 1), and each of them However, the effect is not very ideal since the sideslip
can operate independently. cannot be completely eliminated, and it is hard to co-
ordinate a roll, turn or to trim in a crosswind landing
situation, etc.
The reconfiguration of the combat flying wing is
achieved through functional redundancy design [10]. It
has redundant control surfaces around three axes of roll,
pitch and yaw. Such a redundancy mode guarantees the
flight control reconfiguration capability of the aircraft.
This is one of the significant reasons that modern ad-
vanced aircraft adopt the deployment scheme of con-
trol surfaces with multiple functional redundancy [10].
Fig. 1 Control surface arrangements of conventional air-
craft and flying wing. When a main control surface of an axis fails, depend-
ing on that redundant control surfaces supply certain
Inboard elevons (IEs) and outboard elevons (OEs) control of this axis, the flight safety of the aircraft can
are used for roll or pitch control [3], and their position be guaranteed. Note that all the control surfaces in-
limits are 30°-30°. Split drag rudders (SDRs) and stalled on the flying wing have multi-axis control cou-
spoiler slot deflectors (SSDs) are drag rudders used for pling effects. As a result, when these control surfaces
yaw control with coupling effects in roll and pitch [3-4], fail, they may cause some adverse influences that
and their position limits are 0°-60°. Rate limits of the should be compensated for by other control surfaces.
eight control surfaces are 80-80 (°)/s. For example, when one side SDR of the typical combat
For the flying wing, due to the characteristics of de- flying wing encounters a LIP failure, it is impossible to
graded multi-axis stability [5], multiple control surface be handled anymore. Then the other SDR and both
arrangement and strong nonlinear dynamic feature, a SSDs should undertake the full yaw control missions,
attitude track flight control system [6] with control al- and the elevons need to be deflected to balance the
location is designed based on the theory of nonlinear coupling moments in roll and pitch induced by the LIP
dynamic inversion [7]. Its overall architecture is shown failure of this SDR.
in Fig. 2. Hence, the combat flying wing differs much from
the conventional aircraft in reconfiguration when con-
trol surface failures occur, as shown in Table 1.
sideration of too many failure types, sliding mode con- constrained to ILIP; kė0° means a LOE control sur-
trol is too conservative. Hence, the approaches men- face deflects back to 0°; gk(k)=0 denotes a float con-
tioned above cannot provide good effects in the recon- trol surface does not generate any effective control
figurable flight control design of the flying wing. moments, and k=0° means its deflection is constrained
For the flying wing, depending on the redundant ar- to 0°.
rangement problem of control surfaces, a control allo-
cation block is already introduced into its dynamic 4.2.1. Principle of control allocation
inversion flight control law. In order to realize the
flight control reconfiguration during control surface The control allocation block uses certain optimiza-
failures, control allocation does not need to modify the tion calculations to allocate the three-axis control mo-
original flight control law of the combat flying wing, ments needed by the aircraft motions to relative control
whereas it just needs to appropriately adjust the con- surfaces [8].
straint conditions of the control allocation block. Firstly, the control allocation block has position and
Therefore, this approach can easily and efficiently ful- rate limits for control surfaces, and transforms the rate
fill the reconfigurable flight control design for the fly- limits to the position limits in every calculation cy-
ing wing in the presence of control surface failures. So cle [21]:
it is a comparative better approach suitable for solving
the reconfiguration problem of such aircraft types. °Gi min max(Gi (t 't ) Gi min 't , Gi min )
® (i 1, 2,", n)
°̄Gi max min(Gi (t 't ) Gi max 't , Gi max )
4.2. Implement of flight control reconfiguration ba- (1)
sed on control allocation
where t denotes the calculation time step, G i min and
This reconfigurable design approach uses the control G i max denote respectively the minimum and maximum
allocation block of the dynamic inversion flight control
system for the flying wing to fulfill flight control re- allowable deflections of the ith control surface at the
configuration during control surface failures, combined time instant of t, i(tt) denotes its deflection at the
with failure detection [20] and management system. last time instant, Gi min and Gi max denote respectively
Control law and control allocation block are designed the minimum and maximum deflection rates, and
independently. According to the command of the air- G i min and G i max , the minimum and maximum deflec-
craft motion, the control law obtains the desired control
tion limitations respectively.
moments not affected by control surface failures, and
So, in the case of no failures, the constraint condi-
the control allocation block uses an approach planed
tions of the control allocation block are described as
before to allocate the moments among the operable
n
control surfaces. As a result, such a reconfigurable de-
sign based on control allocation does not need to mod- ¦ gi (Gi ) Md , Gi min gGi gGi max
(2)
i 1
ify the structure and the parameters of the control law,
that it can be directly applied to the flight control sys- where gi(i) denotes the control moment vector gener-
tem design of the combat flying wing. Figure 4 shows ated by the ith control surface, and Md denotes the de-
the basic principle of the reconfigurable flight control sired control moment vector, which consists of roll,
design based on control allocation in the case of con- pitch and yaw control moments, non-dimensional as
trol surface failures. [Cl Cm Cn]T. The first formula represents the
command that control surfaces fulfill the desired con-
trol moments. The second formula is the deflection
constraints of control surfaces.
Secondly, the control allocation block takes into ac-
count several kinds of optimization objectives, such as
minimal total deflection energy, radar cross section, etc.
This paper considers the optimization objective of the
maximal additional drag generated by control sur-
faces [8]:
n
min J ¦ 'CDi (Gi ) (3)
i 1
4.2.2. Adjust strategies of control reconfiguration After the LOE control surface deflects back to the
neutral position, it will not deflect anymore and always
It is assumed that the control surface failure detec- keeps at the deflection of 0°.
tion system can work well, which means that it can 3) Float failure
detect the work state of every control surface quickly When the kth control surface has a failure of float, it
and exactly. Assuming that the kth control surface fails will not generate any effective control moments any-
sometime, for the three typical failures of LIP, LOE more, so its control effectiveness is 0. Now, the float
and float, implements of the reconfigurable flight con- control surface should be isolated immediately so as
trol design based on control allocation are discussed not to take part in control allocation anymore.
below, i.e., the adjust strategy of constraint conditions In the control allocation block, the deflection posi-
(Eq. (2)) in the control allocation block after control tion of the float control surface is constrained to 0°.
surface failures. The constraint conditions are adjusted as
1) LIP failure
k 1 n
°¦ gi (Gi ) 0 u gk (G k ) ¦ gi (Gi )
When the kth control surface has a LIP failure, it Md
will keep at the same deflection of LIP, and it cannot °i 1 i k 1
respond to any deflection commands. Now, its deflec- ® (6)
tion rate is 0 (°)/s. °Gi min gGi gGi max (i z k )
°G 0q (i k )
In the control allocation block, the deflection posi- ¯ i
tion of this control surface is constrained to the LIP
deflection. The constraint conditions are adjusted as where 0×gk(k) denotes that the float control surface
does not generate effective control moments, the first
n
°¦ gi (Gi ) Md
formula represents the command that control surfaces
°i 1 fulfill the desired control moments, the second one is
® (4) the deflection constraints of normal control surfaces,
°Gi min gGi gGi max (i z k ) and the last one represents that the deflection of the
°G I (i k )
¯ i LIP faulty control surface is constrained to 0°.
where the first formula represents the command that In fact, the deflection of the float control surface is
control surfaces fulfill the desired control moments, the not 0°. In the calculation and simulation, that the de-
second one is the deflection constraints of normal con- flection of this control surface is appointed 0° repre-
trol surfaces, and the last one represents that the de- sents that its control effectiveness is 0.
flection of the LIP control surface is constrained to Take the typical combat flying wing for example. At
I LIP. the time instant of tt, the deflections of its eight con-
The allocation result of the faulty control surface trol surfaces are 2°, 30°, 28°, 23°, 36°, 5°, 60° and
will always be the deflection of LIP. 60°. Assuming that the right SDR fails and is detected
2) LOE failure at the time instant of t, the non-dimensional desired
Control effectiveness of the kth control surface de- control moment vector is [0.01 0.01 0.01]T, and
creases, when it is in the failure of LOE. Though this the calculation step t is 0.02 s. Table 2 gives the allo-
control surface still has certain control power, based on cation results when the right SDR is in three typical
the conservative consideration of avoiding aggravating failure conditions.
its damage degree, it should deflect back fast to the
Table 2 Allocation results after the right SDR is sub-
neutral position. jected to three typical failures
In the control allocation block, the constraint condi-
tions are adjusted as Allocation results of control surfaces/(°)
Generate
Failure
k1 n IE OE SDR SSD desired
°¦ gi (Gi ) (1 cLOE ) u gk (Gk ) ¦ gi (Gi ) Md
type
moments
Left Right Left Right Left Right Left Right
°i 1 i k 1
°G gG gG (i z k) Normal 3.3 30 29.6 22.2 37.6 5.6 60 60
° i min i i max
® LIP 3.3 30 29.6 22.3 37.3 5.0 60 60
° Gi (t ' t) Gi min' t Gi (t ' t) ! Gi max' t Yes
°G °®0q
LOE 3.4 30 29.6 22.6 36.4 3.4 60 60
Gi min' tgGi (t ' t)gGi max' t (i k)
°i ° Float 3.6 30 29.6 23.1 34.4 0 60 60
°¯ ¯Gi (t ' t) Gi max' t Gi (t ' t) Gi min' t
(5) As can be seen in Table 2, during the three typical
where cLOE denotes the damage ratio of the LOE con- control surface failures of the right SDR, the flight
trol surface, the first formula represents the command control reconfiguration strategy comparatively better
that control surfaces fulfill the desired control moments, assure that the desired control moments are fulfilled
the second one is the deflection constraints of normal precisely. Besides, the allocation results of the faulty
control surfaces, and the third to the last formulae de- control surface correspond with the adjust strategies:
scribe the process that the faulty control surface de- allocated to 5° after a LIP failure, 3.4° after a LOE
flects back to 0°. failure, or 0° after a float failure.
· 498 · WANG Lei et al. / Chinese Journal of Aeronautics 25(2012) 493-499 No.4
good.
When the flight control system has no reconfigura-
tion, it continues to allocate control moments to both
the SSDs. But actually, they cannot provide necessary
moments anymore. Next to that, the errors of the flight
control system are accumulated, and then some control
surfaces gradually saturate at their position limits (e.g.
the left IE and the left OE, see Fig. 6). Finally, the con-
trol surfaces cannot provide sufficient control moments.
However, when the flight control system has recon-
figuration, it allocates desired control moments to the
six available control surfaces, therefore the errors of
moment allocations decrease. So if the remaining con-
trol power is adequate, the desired control moments
can be allocated comparatively accurately, also the
aircraft can comparatively better perform the flight
mission.
6. Conclusions
via appropriately modifying the control allocation cal University Press, 2007; 23-24. [in Chinese]
block, and then it can comparatively effectively realize [11] Durham W C. Constrained control allocation. AIAA-
the reconfigurable design during control surface fail- 1992-4550, 1992.
ures. So it is suitable for the modular design of the [12] Xiao Q G, Zhang M, Hu S S. Adaptive reconfiguration
control for fighters based on weighted multi-
complicate flight control system for the flying wing. ple-model-structure. Transactions of Nanjing Univer-
3) Despite the presence of control surface failures, sity of Aeronautics & Astronautics 2010; 27(3):
the combat flying wing using this flight control recon- 219-225. [in Chinese]
figuration approach can still guarantee its flight safety, [13] Li W Q, Chen Z J. Control law reconfiguration under
and can comparatively better perform some flight mis- control surface failure for nonlinear aircraft dynamic.
sions with certain amplitude. Journal of Beijing University of Aeronautics and As-
tronautics 2003; 29(5): 428-433. [in Chinese]
[14] Bodson M. A reconfigurable nonlinear autopilot.
AIAA-2002-4549, 2002.
[15] Brinker J, Wise K A. Flight testing of a reconfigurable
References flight control law of the X-36 tailless fighter aircraft.
AIAA-2000-3941, 2000.
[16] Wells S R, Hess R A. MIMO sliding mode control for a
[1] Boskovic J D, Mehra R K. Robust fault-tolerant control tailless flight aircraft, an alternative to reconfigurable
design for aircraft under state-dependent disturbances. architectures. AIAA-2002-4650, 2002.
AIAA-2003-5490, 2003. [17] Davidson J B, Lallman F J, Bundick W T. Integrated
[2] Bowlus J A, Multhopp D, Banda S S. Challenges and reconfigurable control allocation. AIAA-2001-4083,
opportunities in tailless aircraft stability and control. 2001.
AIAA-1997-3830, 1997. [18] Shertzer R H, Zimpfer D J, Brown P D. Control alloca-
[3] Ma C, Li L, Wang L X. Design method of controllabil- tion for the next generation of entry vehicles. AIAA-
ity of low aspect-ratio flying wing configuration com- 2002-4849, 2002.
bat aircraft. Acta Aeronautica et Astronautica Sinica [19] Boskovic J D, Mahra R K. A multiple model-based
2008; 29(4): 788-794. [in Chinese] reconfigurable flight control system design. Proceed-
[4] Wang L, Wang L X, Jia Z R. Control features and ap- ings of the 37th IEEE Conference on Decision & Con-
plication characteristics of split drag rudder utilized by trol. 1998; 4503-4508.
flying wing. Acta Aeronautica et Astronautica Sinica [20] Huang C T, Wang L X. On-line fault diagnosis for con-
2011; 32(8): 1392-1399. [in Chinese] trol surfaces of multi-control effector flying wings.
[5] Li L, Ma C, Wang L X. Stability features of low as- Acta Aeronautica et Astronautica 2011; 32(1): 58-66.
pect-ratio flying wings. Acta Aeronautica et Astronau- [in Chinese]
tica Sinica 2007; 28(6): 1312-1317. [in Chinese] [21] Durham W C, Leedy J Q. Real-time evaluation of con-
[6] Liu J R, Shen G Z. Attitude control based on inverse trol allocation with rate limiting. AIAA-1998-4460,
dynamics and online parameter identification. Journal 1998.
of Beijing University of Aeronautics and Astronautics
2005; 31(2): 111-115. [in Chinese]
[7] Ito D, Georgie J, Valasek J, et al. Reentry vehicle flight Biographies:
controls design guidelines: dynamic inversion. NASA/
TP-2002-210771, 2002.
WANG Lei received B.S. degree from Beihang University in
[8] Wang L, Wang L X, Jia Z R. Control allocation method
2008, and is now a Ph.D. candidate. His main research inter-
for combat flying wing with multiple control surfaces.
ests are aircraft flight dynamics and control.
Acta Aeronautica et Astronautica Sinica 2011; 32(4):
E-mail: wolsur@163.com
571-579. [in Chinese]
[9] Clark R N, Frank P M, Patton R J. Fault diagnosis in
dynamic systems: theory and application. New York: WANG Lixin is a professor in Beihang University. His main
Prentice Hall, 1989; 134. research interests lie in aircraft design, flight dynamics and
[10] Yang W, Zhang L G, Yang C X, et al. Fault tolerant flight control.
flight control system. Xi’an: Northwestern Polytechni- E-mail: bhu_wlx@tom.com